Embed Size (px)
Citation preview
Department of Organic and Macromolecular Chemistry
Research group NMR and Structure Analysis
Effects of fluorination on the conformation of
proline: an NMR study
Thesis submitted to obtain
the degree of Master of Science in Chemistry by
Emile OTTOY
Academic year 2015 - 2016
Promoter: dr. Davy Sinnaeve
Copromoter: prof. dr. José Martins
Department of Organic and Macromolecular Chemistry
Research group NMR and Structure Analysis
Effects of fluorination on the conformation of
proline: an NMR study
Thesis submitted to obtain
the degree of Master of Science in Chemistry by
Emile OTTOY
Academic year 2015 - 2016
Promoter: dr. Davy Sinnaeve
Copromoter: prof. dr. José Martins
i
List of abbreviations Ac-Pro-OMe Methyl acetyl-L-prolinate Ac-(4R)-FPro-OMe Methyl acetyl-L-(4R)-fluoroprolinate Ac-(4S)-FPro-OMe Methyl acetyl-L-(4S)-fluoroprolinate Ac-4,4-F2Pro-OMe Methyl acetyl-L-4,4-difluoroprolinate CD Circular dichroism COSY Correlation spectroscopy DFT Density functional theory EXSY Exchange spectroscopy Flp (4R)-fluoroproline flp (4S)-fluoroproline FRET Förster resonance energy transfer GUI Graphical user interface HMBC Heteronuclear multiple bond correlation HOESY Heteronuclear Overhauser enhancement spectroscopy HSQC Heteronuclear single quantum correlation NMR Nuclear magnetic resonance nOe Nuclear Overhauser enhancement NOESY Nuclear Overhauser enhancement spectroscopy PPI Polyproline I helix PPII Polyproline II helix Pro Proline PRR Proline rich region PSYCHE Pure shift yield by chirp excitation PSYCHEDELIC PSYCHE to deliver individual couplings RMSD Root mean squared difference TOCSY Total correlation spectroscopy Xaa Any amino acid
ii
Dankwoord
Alvorens deze thesis te beginnen zou ik graag de tijd nemen om een aantal mensen
te bedanken, te beginnen met de mensen, zonder wiens inbreng, deze thesis niet
mogelijk was geweest. Mijn dank gaat eerst en vooral uit naar Davy, om zo een goede
begeleider te zijn en me ten allen tijde met raad en daad bij te staan. Jouw
enthousiasme werkt zeer aanstekelijk en ik ben blij dat ik het fluoroproline team mogen
versterken heb. Ook professor José Martins zou ik willen bedanken om me de kans te
geven om mijn thesis in zijn groep te doen. Voorts zou ik ook Gert-Jan willen bedanken
voor de synthese van de onderzochte moleculen, die anders niet onderzocht hadden
kunnen worden. Mijn dank gaat tevens uit naar Krisztina, om me vertrouwd te maken
met de spectrometers, en naar Ilya Kuprov, voor het ter beschikking stellen van zijn
computationele berekeningen.
Bedankt aan alle leden van de NMR groep: Tim, voor de technische hulp en voor me
op te lappen als dat nodig was; Matthias en Dieter, voor me een aantal
labohandelingen bij te brengen; Niels, voor antwoord te geven op mijn (waarschijnlijk
pietluttige) vragen; en Kim, Jonathan en Emile, voor de sfeer en gezelligheid. Jullie
humor, vriendelijkheid en professionaliteit zorgen ervoor dat ik graag naar “mijn
bureau” kwam.
Omdat de thesis ook het sluitstuk is van deze vijfjarige opleiding, zou ik graag ook van
de gelegenheid gebruikt willen maken om een aantal van mijn medestudenten de
bedanken. Eerst en vooral, Olieux, Matthi, Hannes en Laurent, voor de vijf
onvergetelijk jaren en de vele epische kaartnamiddagen. Een betere vriendengroep
kan men zich niet wensen. Ook bedankt Yann, om mijn persoonlijke taxichauffeur te
spelen het laatste half jaar. En als laatste, maar zeker niet de minste, Jurgen en Anne-
Mare, zonder wiens leuke gezelschap het maar een eenzaam thesisjaar zou geweest
zijn.
Als laatste wil ik nog graag mijn ouders bedanken, om me de kans te geven deze
opleiding aan te vatten, en om me steeds te steunen ook in de moeilijke tijden.
iii
Table of contents List of abbreviations………………………………………………………………… i Acknowledgements………………………………………………………………… ii Dutch summary……………………………………………………………………... v Scientific article……………………………………………………………………... vii 1. Introduction………………………………………………………………………. 1 1.1. Proline…………………………………………………………………... 1 1.2. Organofluorine chemistry…………………………………………….. 6 1.3. Applications of fluorine in protein sciences…………………………. 8 2. Experimental methods………………………………………………………….. 13 2.1. Introductory information………………………………………………. 13 2.1.1. The scalar coupling…………………………………………. 13 2.1.2. Studied samples…………………………………………….. 13 2.2. Assignment of the spectra……………………………………………. 14 2.2.1. General features of the spectra……………………………. 14 2.2.2. General assignment strategy of the 1H spectra………….. 15 2.2.2.1. Some special cases………………………………. 19 2.2.2.1.1. The Ac-4,4-F2Pro-OMe samples……… 19 2.2.2.1.2. Overlap and strong coupling…………... 19 2.2.3. General assignment strategy of the 13C spectra…………. 21 2.2.4. General assignment strategy of the 19F spectra…………. 22 2.3. Measuring coupling constants……………………………………….. 22 2.3.1. Measuring 1H-1H couplings………………………………… 23 2.3.1.1. PSYCHEDELIC…………………………………… 23 2.3.1.2. Encountered difficulties…………………………... 25 2.3.2. Measuring 1H-19F couplings………………………………... 26 2.4. NMR techniques for measuring relaxation and kinetics…………… 28 2.4.1. Inversion recovery…………………………………………... 28 2.4.2. Selective inversion recovery……………………………….. 28 2.4.3. 2D EXSY……………………………………………………... 29 3. Puckering analysis………………………………………………………………. 30 3.1. Mathematical description of five-ring conformations………………. 30 3.2. The relationship between torsion angles and scalar couplings…... 31 3.3. Fitting conformations………………………………………………….. 34 3.3.1. MatLab script………………………………………………… 35 3.3.2. Approximations and limitations…………………………….. 36 3.4. Results and discussion……………………………………………….. 37 3.4.1. Ac-Pro-OMe………………………………………………….. 37 3.4.2. Ac-(4R)-FPro-OMe………………………………………….. 40 3.4.3. Ac-(4S)-FPro-OMe………………………………………….. 42 3.4.4. Ac-4,4-F2Pro-OMe…………………………………………... 43 3.4.5. Fitting with the Barfield correction…………………………. 46 4. Thermodynamics and kinetics of the cis/trans equilibrium………………….. 47 4.1. Investigation of the thermodynamics………………………………... 47 4.1.1. Theory………………………………………………………… 47 4.1.2. Results and discussion……………………………………... 48 4.2. Investigation of the kinetics…………………………………………... 50 4.2.1. Theory………………………………………………………… 50 4.2.1.1. Selective inversion recovery equation………….. 51
iv
4.2.1.2. EXSY equations…………………………………… 52 4.2.1.3. Arrhenius and Eyring equations…………………. 53 4.2.2. Results and discussion……………………………………... 54 5. General conclusion……………………………………………………………… 58 5.1. Concluding remarks…………………………………………………… 58 5.2. Future prospects………………………………………………………. 59 6. Supporting information………………………………………………………….. 61 7. References……………………………………………………………………….. 112
v
Samenvatting
NMR is een krachtige techniek voor het ophelderen van structuur en dynamica van
peptiden en eiwitten. Wanneer men evenwel te maken krijgt met prolinerijke regio’s
(PRR’s), ontstaan er moeilijkheden door de afwezigheid van amide protonen in deze
regio’s, die normaal gebruikt worden voor de toekenning via TOCSY. De problemen
die dit met zich meebrengt, zouden verholpen kunnen worden door fluorering van
proline. De aanwezigheid van fluor, dispergeert niet alleen de waarden van de
chemische verschuivingen van het opeengepakte 1H spectrum, maar maakt ook het
gebruik van 19F NMR mogelijk. Als men conformationele analyse van gefluoreerde
prolines doet, dient men er rekening mee te houden, dat de conformatie van proline
sterk beïnvloed zal worden door de stereo-elektronische effecten die fluor uitoefent.
Fluoroprolines kunnen dus ook gebruikt worden om de structuureigenschappen van
PRR’s te moduleren. Twee conformationele evenwichten zijn van belang voor proline,
zijnde de vijfring pucker (letterlijk opplooiing), die voor proline derivaten varieert tussen
Cγ endo en Cγ exo (beiden enveloppe conformaties waarbij Cγ uit het vlak ligt), en de
amidebinding cis/trans isomerie. Een grondige conformationele analyse van alle
mogelijke fluoroproline derivaten is dus nodig alvorens men kan overgaan tot de
incorporatie van fluoroprolines in peptiden of eiwitten.
Om zulk een grondige conformationele analyse te kunnen uitvoeren, werd recent een
nieuwe NMR methodologie ontwikkeld of basis van pure shift, zijnde PSYCHEDELIC.
PSYCHEDELIC laat ons toe om alle individuele 1H-1H scalaire koppelingen naar een
geselecteerd proton, gemakkelijk te meten als doubletten in een 2D J-geresolveerd
spectrum. Deze koppelingen kunnen vervolgens met behulp van de zogenaamde
Karplus curves gerelateerd worden aan torsiehoeken, die op hun beurt nodig zijn om
de ring pucker te bepalen.
Het doel van deze thesis was, om voor een aantal gekende (in termen van
conformationele voorkeuren) fluoroproline derivaten, zijnde Ac-Pro-OMe, Ac-(4R)-
FPro-OMe, Ac-(4S)-FPro-OMe en Ac-4,4-F2Pro-OMe, het potentieel van
PSYCHEDELIC te onderzoeken, d.m.v. zo veel mogelijk gemeten 1H-1H en 1H-19F
koppelingen een uitgebreide analyse van de ring pucker uit te voeren, en om de
gerapporteerde conformationele voorkeuren in de literatuur de bevestigen. Ook de
cis/trans isomerie van de amide binding werd heronderzocht.
vi
Het merendeel van de scalaire koppelingen kan zonder probleem met behulp van
PSYCHEDELIC gemeten worden. Echter een probleem ontstaat wanneer protonen
sterk gekoppeld zijn. Twee protonen zijn sterk gekoppeld als hun koppelingsconstante
van dezelfde grootteorde is als hun resonantiefrequentieverschil. Deze sterke
koppelingen leiden tot artefacten die de interpretatie van de spectra zodanig
bemoeilijken, dat het meten van individuele koppelingen vaak niet meer mogelijk is. In
sommige gevallen was het toch nog mogelijk om sommen of gemiddeldes van
koppelingen van de sterk gekoppelde protonen te meten.
Met de puckering analyse op basis van gemeten scalaire koppelingen en 1H-1H en 1H-
19F Karplus relaties, was het mogelijk om de in de literatuur gerapporteerde resultaten
te bevestigen voor drie van de vier onderzochte moleculen. Ac-Pro-OMe vertoonde
een 7:3 evenwicht tussen Cγ endo en Cγ exo respectievelijk, het evenwicht in Ac-(4R)-
FPro-OMe is sterk verschoven richting Cγ exo en het evenwicht in Ac-(4S)-FPro-OMe
is sterk verschoven richting Cγ endo. Alleen voor Ac-4,4-F2Pro-OMe kon niet besloten
worden dat het puckering evenwicht gelijkaardig is aan dat van Ac-Pro-OMe. Om dit
te kunnen doen, en om het mogelijk bestaan van een kleine fractie ander conformeer
voor de gemonofluoreerde moleculen te kunnen bewijzen, zal een verdere analyse en
verfijning van de 1H-1H en vooral 1H-19F Karplus relaties in de context van
fluoroprolines nodig zijn, aangezien de betrouwbaarheid van deze relatie onbekend is.
Het merendeel van de bekomen thermodynamische en kinetische resultaten van de
cis/trans isomerie zijn in overeenstemming met deze uit de literatuur en kunnen tevens
gerationaliseerd worden. De grootste portie aan toekomstig werk op dit vlak zal gewijd
worden aan het uitbreiden van de temperatuursafhankelijke datapunten zodat er een
preciezere uitspraak bekomen kan worden in verband met reactie-enthalpie, reactie-
entropie, activeringsenergie, activeringsenthalpie en activeringsentropie. Zeker voor
beide entropieën, waarop de onzekerheden zo groot zijn dat niet met zekerheid kan
gezegd worden of ze positief of negatief zijn, zou dit de interpretatie kunnen
vergemakkelijken.
vii
Effects of Fluorination on the Conformation of Proline: An NMR Study
E. Ottoya, G. J. Hofmanb, B. Linclaub, J. C. Martinsa, and D. Sinnaevea
a Department of Organic and Macromolecular Chemistry, Ghent University, Ghent, Belgium
b Department of Chemistry, University of Southampton, Southampton, UK
To verify the conformational preferences of four (fluoro)proline
analogues (Ac-Pro-OMe, Ac-(4R)-FPro-OMe, Ac-(4S)-FPro-OMe and
Ac-4,4-F2Pro-OMe) (1), puckering analysis was performed using a
methodology developed by Hendrickx et al. (2), and cis/trans isomerism
was investigated using the strategies from Eberhardt et al. and Bain et
al. (3, 4). For this, the recently developed NMR technique for extracting
scalar couplings, PSYCHEDELIC (5), was used. The majority of vicinal 1H-1H and 1H-19F couplings could be extracted at pure shift resolution.
The results of the puckering analysis could confirm the literature
observations for Ac-Pro-OMe, Ac-(4R)-FPro-OMe and Ac-(4S)-FPro-
OMe, but were inconclusive for Ac-4,4-F2Pro-OMe. Thermodynamic
and kinetic studies of the cis/trans isomerism, yielded similar results as
those reported in literature.
Keywords: fluoroproline, conformational analysis, NMR
Introduction
NMR is a powerful tool for performing site-specific structure and dynamics studies of protein
sequences (6). However, when dealing with proline rich regions (PRRs), this technique is
plagued by the absence of amide protons. A possible solution to this problem is offered by
proline fluorination. The presence of the fluorine atom not only disperses the chemical shift
values of the otherwise crowded proline 1H spectrum, but also makes the use of 19F NMR
possible. The 19F spectra will be easily interpretable if only a limited amount of fluorine atoms
is introduced. However, it needs to be taken into account that fluorine induces strong
stereoelectronic effects that will affect the proline conformation (7). Therefore, it is important
to evaluate the conformational preferences of the incorporated fluoroproline, before studying
the conformation of the whole segment, or before incorporating fluoroproline with the purpose
of modulating the protein structure. A library, containing the conformational preferences of all
possible fluoroproline analogues, would thus be very practical. There are two conformational
equilibria that are of interest: first, the five-membered ring pucker, that is usually assumed to
be an equilibrium between Cγ exo and Cγ endo, both envelope conformations in which the Cγ
atom is out of plane (Figure 2), and second, the peptide bond cis/trans equilibrium (8). In order
to perform an extensive analysis of the puckering equilibrium, new advanced NMR
methodology for measuring coupling constants, based on pure shift NMR, was employed (9).
The purpose was to evaluate this methodology by reinvestigating the puckering of already
studied fluoroproline analogues, being Ac-Pro-OMe, Ac-(4R)-FPro-OMe, Ac-(4S)-FPro-OMe
and Ac-4,4-F2Pro-OMe (see Figure 1). Also a reinvestigation of the cis/trans equilibrium was
performed for comparison with reported values.
viii
Figure 1. The investigated molecules. From left to right: Ac-Pro-OMe, Ac-(4R)-FPro-OMe,
Ac-(4S)-FPro-OMe and Ac-4,4-F2Pro-OMe.
Figure 2. L-Proline. Left: Cγ endo pucker. Right: Cγ exo pucker.
Experimental
All molecules were studied in CDCl3 (Euriso-Top, 99.8% D) and D2O (Euriso-Top 99.9% D),
at concentrations of approximately 20 mM. All NMR measurements were performed on either
one of four different spectrometers: a Bruker Avance II spectrometer equipped with either a
5mm 1H-13C-31P TXI-Z probe or a 5mm 1H-13C-19F TXO-Z probe, operating at 1H, 13C and 19F
frequencies of 500.13 MHz, 125.76 MHz and 470.59 MHz respectively; a Bruker Avance III
spectrometer equipped with a 5mm BBI-Z probe, operating at 1H and 13C frequency of
500.13 MHz and 125.76 MHz respectively; a Bruker Avance II spectrometer equipped with a
5mm 1H-13C-31P TXI-Z probe, operating at 1H and 13C frequency of 700.13 MHz and
176.05 MHz respectively; or a Bruker Avance II spectrometer equipped with a 5mm 1H-BB-19F TBI-Z probe, operating at 1H and 19F frequency of 400.13 MHz and 376.50 MHz
respectively. All measurements were performed at 298.0 K unless mentioned otherwise.
Spectral assignment was performed using 1D 1H, 1D 19F, 2D 1H-1H COSY, 2D 1H-13C HSQC,
2D 1H-13C HMBC, 2D 1H-1H NOESY and 2D 1H-19F HOESY spectra. The main NMR
technique used for measuring coupling constants was PSYCHEDELIC (Pure Shift Yielded by
CHirp Excitation to DELiver Individual Couplings) (5). In this experiment, which is based on
the PSYCHE pure shift experiment (10), a 1H is selected, and generates a 2D J-resolved
spectrum that contains only the couplings to this 1H as simple doublets. In the 2DJ spectrum,
the doublets are dispersed along a –45° pattern. This can be tilted by 45° to align the splitting
fully along the F1 dimension, leaving only chemical shift information along F2, and providing
a resolution similar to the PSYCHE pure shift experiment. One can excite multiple 1H’s at the
same time, but to retain the simple doublet splittings, and thus unambiguous interpretation,
these 1H’s may not have mutual coupling partners amongst the 1H’s to which the couplings are
measured. If only one fluorine atom was present in the molecule, 1H-19F couplings could be
measured using a 1D PSYCHE experiment (10). In this experiment, 1H-1H splittings have been
eliminated, meaning only the 1H-19F couplings, will be present as doublets and thus easily
measurable. For Ac-4,4-F2Pro-OMe, a variant of PSYCHEDELIC was used that measures 1H-19F couplings in a similar way as 1H-1H couplings.
The extracted couplings can subsequently be used in the puckering analysis, which uses the
Altona-Sundaralingam formalism (11) (equation [1]) for the description of five-membered ring
ix
conformations based on endocyclic torsion angles (νi). The fitting program starts with initial
guesses of puckering parameters P and νmax, equation [1] is used to calculate the endocyclic
torsion angles, these are converted to exocyclic torsion angles by adding 0° for cisoid couplings
or by adding ± 120° for transoid couplings, which are then fed into the 1H-1H or 1H-19F Karplus
relation (equations [2] and [3], coefficients are calculated as described in (12, 13), input
parameters can be found in the supporting information) to yield calculated coupling values.
When fitting two conformations, the initial guess of the minor fraction is used to calculate the
weighted average scalar couplings, which are then compared to the experimental couplings.
Goodness-of-fit is determined by the RMSD value (equation [4]), which is checked for
convergence. Until convergence is reached, the described fitting routine is iterated (2).
Ktrans/cis was calculated by taking the ratio of the integral of a certain resonance (methyl 1H or 19F) of the trans form and the integral of the same resonance of the cis form. This was done
using quantitative 1D spectra. For determining kex (= k1 + k-1), either a selective inversion
recovery or a 2D EXSY experiment was performed. In the selective inversion recovery a 90°x
– Δ – 90°-x pulse sequence is applied, with Δ half of the inverse frequency difference between
the involved resonances (cis and trans of a certain resonance). The off-resonance signal will be
inverted, while the on-resonance signal will remain unaffected. By varying the recovery delays,
the intensity change of the on-resonance signal (due to exchange with the inverted signal) can
be followed. These data points are then fitted to equation [5], in which T1 (determined via a
regular inversion recovery experiment) serves as input and which yields kex. In a 2D EXSY,
cross-peak and diagonal peak volumes of the exchanging spins (I and S), together with the
mixing time, are used to calculate kex via equation [6], in which rI is the ratio of the diagonal
peak volume over the cross-peak volume (both for resonance I) and pI and pS are the equilibrium
populations of spins I and S respectively. For both fitting procedures (puckering and kex), 95%
uncertainties were calculated using a Monte Carlo analysis (14).
𝜈𝑖 = 𝜈𝑚𝑎𝑥 cos (𝑃 +
4𝜋
5𝑖)
[1]
𝐽𝐻𝐶𝐶𝐻 = 𝐶0 + 𝐶1 cos(𝜑) + 𝐶2 cos(2𝜑) + 𝐶3 cos(3𝜑) + 𝑆2sin (2𝜑) [2]
𝐽𝐻𝐶𝐶𝐹 = 40.61cos2(𝜙) − 4.22 cos(𝜙) + 5.88
+∑𝜆𝑖[−1.27 − 6.20cos2(𝜉𝑖𝜙 + 0.20𝜆𝑖)]
𝑖
− 3.72 (𝑎𝐹𝐶𝐶 + 𝑎𝐻𝐶𝐶
2− 110) cos2(𝜙)
[3]
𝑅𝑀𝑆𝐷 = √1
𝑁∑(𝐽𝑖
𝑐𝑎𝑙𝑐 − 𝐽𝑖𝑒𝑥𝑝𝑡
)2𝑁
𝑖
[4]
𝐼(𝑡) = 𝐼0 (1 − (𝑎𝑒
−𝑡𝑇1 − 𝑏𝑒
−𝑡(1𝑇1+𝑘𝑒𝑥)))
[5]
x
𝑘𝑒𝑥 = −
ln(1 −
𝑝𝐼𝑝𝑆𝑟𝐼
1 + 𝑟𝐼)
𝑡𝑚
[6]
Results and Discussion
Measuring couplings
The measured vicinal 1H-1H and 1H-19F couplings can be found in Tables 7, 8, 9 and 10 in
the supporting information. With PSYCHEDELIC, it was possible to measure the majority of
couplings without any problems. The couplings developed as clearly resolved doublets and the
precision is approximately 0.1 Hz. In some cases, the couplings were smaller than the linewidth
of a single line (approximately 1 Hz), which means that no doublet was visible. Sometimes this
could be overcome by applying Lorentz-to-Gauss resolution enhancement (‘line broadening
parameter’ in the supporting information), but this was not always successful. In those cases, it
could only be reported that the coupling has an upper limit of 1 Hz. A more serious issue occurs
when protons are strongly coupled, i.e., when their mutual coupling constant is of the same
order of magnitude as their resonance frequency difference. Protons that are strongly coupled
not only typically feature overlap between their multiplets, which makes selective excitation of
the separate protons difficult, but it also leads to a series of artefacts, that complicate spectral
interpretation. In some cases (cis Ac-Pro-OMe in CDCl3, trans Ac-(4R)-FPro-OMe in CDCl3,
trans Ac-(4S)-FPro-OMe in D2O), it was still possible to measure sums or averages of couplings
for protons involved in strong coupling, but these results are less useful compared to knowing
their individual couplings. In case of trans Ac-Pro-OMe in D2O, 8 out of 10 possible vicinal 1H-1H couplings could not be measured. Sometimes, overlap hampered the unambiguous
measurement of 1H-19F couplings in the 1D PSYCHE pure shift spectrum. Sometimes it was
possible to measure these couplings in the 2D 1H-13C HSQC as the signals in question were
resolved along the 13C dimension.
Puckering analysis
Only the results of the samples where (nearly) all of the 1H-1H and 1H-19F couplings could
be measured are discussed here, as these are the most reliable. The puckering parameters can
be understood in the following manner. P indicates which atoms are out of plane, while νmax
indicates how far these atoms are tilted out of plane. A Cγ exo pucker has a P value of 18°,
while a Cγ endo pucker has a P value of 198°. The puckering parameters can be most
conveniently presented in a polar plot, called a pseudorotation wheel, in which P is plotted on
the polar axis and νmax on the radial axis. A selection of such pseudorotation wheels can be
found in Figure 3.
Ac-Pro-OMe. According to literature (1), the ring pucker in Ac-Pro-OMe is predominantly
Cγ endo with a minor fraction Cγ exo (30%). Fitting one conformer for Ac-Pro-OMe yields
results with very high uncertainties and RMSD values that indicate that it is not possible to
adequately explain the observed couplings based on one conformation only. The results from
fitting two conformations for Ac-Pro-OMe, which yielded much lower RMSD values, can be
found in Table I. These results confirm the observations from literature, as the minor fraction
(1 in this case) corresponds to a Cγ exo pucker and the major to a Cγ endo pucker. For the trans
conformer, the same fractions as reported in literature are obtained, but for the cis conformer,
the fraction of Cγ exo is only about half of the reported value. This can be explained when
xi
bearing in mind that the puckering equilibrium and cis/trans equilibrium are coupled due to n
→ π* interactions (15). In an n → π* interaction, the lone pair (n) electron orbital of one
carbonyl oxygen overlaps with the π* orbital of another carbonyl group, providing an
electronically stabilizing effect. These interactions can only occur in the trans conformer, and
the Cγ exo pucker preorganizes the backbone in such a way that these interactions are favoured,
while the Cγ endo pucker does not (16). Therefore, a trans conformer will have a preference for
a Cγ exo pucker and vice versa. TABLE I. Results of the puckering analysis of all samples for which (nearly) all individual couplings could be measured.
Results for Ac-(4R)-FPro-OMe and Ac-(4S)-FPro-OMe listed here, are without use of 1H-19F couplings.
Ac-[X]-OMe RMSD (Hz) P1 (°) νmax,1 (°) x1 (%) P2 (°) νmax,2 (°)
Pro trans CDCl3 0.54 17.1 ± 7.1 49.7 ± 17.5 29.5 ± 6.2 187.9 ± 5.9 31.5 ± 7.5
Pro cis D2O 0.81 13.4 ± 10.2 43.7 ± 20.5 16.9 ± 4.8 182.8 ± 3.7 36.4 ± 4.4
(4R)FPro trans CDCl3 0.94 24.9 ± 2.1 48.0 ± 2.0 / / /
(4R)FPro cis CDCl3 0.37 24.1 ± 1.7 40.7 ± 1.1 / / /
(4R)FPro trans D2O 0.44 13.5 ± 1.7 43.5 ± 0.9 / / /
(4R)FPro cis D2O 0.45 25.0 ± 1.7 41.1 ± 1.2 / / /
(4S)FPro trans CDCl3 0.74 206.5 ± 2.9 37.1 ± 1.1 / / /
(4S)FPro cis CDCl3 0.68 199.3 ± 3.0 38.0 ± 1.1 / / /
(4S)FPro cis D2O 0.76 203.2 ± 2.6 41.6 ± 1.2 / / /
Ac-(4R)-FPro-OMe. The result listed in Table I is based on 1H-1H couplings alone, and
assumes a single conformation. According to literature (1), the ring pucker of Ac-(4R)-FPro-
OMe has a large preference for Cγ exo. All the fitted P values for this molecule are indeed close
to 18°, while the RMSD values are quite low. This not only confirms that Cγ exo pucker is
dominant in Ac-(4R)-FPro-OMe, but this also suggests that it is reasonable to assume only one
ring pucker is adopted. When including 1H-19F couplings, the outcome did not change
significantly. If one wants to fit two conformations, inclusion of the 1H-19F couplings is
unavoidable, as the maximum number of available vicinal 1H-1H couplings (six) is insufficient
to fit five unknown parameters. For the two trans conformers, the resulting uncertainties on the
fraction were unreasonably high, so that it could not be proven that a second fraction is present.
For the two cis conformers, uncertainties on the fractions were acceptable, and the major pucker
coincides with the Cγ exo pucker. However, the puckering parameters of the minor pucker are
unrealistic, leading to endocyclic torsion angles of over 100°. It is not surprising that attempting
to obtain two conformers is less reliable in this case, since a very minor fraction conformer will
be very sensitive to small coupling measurement errors and the limited accuracy of the Karplus
relations. It is very likely that there exists an equilibrium between a Cγ exo and a Cγ endo form
with a very small fraction of the latter, but this cannot be proven with the current methodology.
For this, proper validation of the 1H-19F Karplus relation, used in the context of fluoroprolines,
is in order.
Ac-(4S)-FPro-OMe. The result listed in Table I is based on 1H-1H couplings alone, and
assumes a single conformation. In literature (1), Cγ endo is the observed dominant pucker for
Ac-(4S)-FPro-OMe. The fitted P values are all close to 198°, so the observations in literature
can again be confirmed. Again, the low RMSD values indicate that assuming one puckering
conformer adequately fits the data. When including the 1H-19F couplings, roughly the same
results were obtained. When fitting two conformations (including 1H-19F couplings, vide
supra), in all cases, the uncertainties on the fractions were very large. An analogous conclusion
as for Ac-(4R)-FPro-OMe can be drawn, that it is not possible to adequately fit two conformers
to the data with the methodology in use.
xii
Figure 3. Selection of pseudorotation wheels. Top left: Two fitted puckers for trans Ac-Pro-
OMe in CDCl3. Top right: One fitted pucker for trans Ac-(4R)-FPro-OMe in D2O without use
of 1H-19F couplings. Bottom left: One fitted pucker for cis Ac-(4S)-FPro-OMe in CDCl3 without
use of 1H-19F couplings. Bottom right: Two fitted puckers for trans Ac-4,4-F2Pro-OMe in
CDCl3 with use of 1H-19F couplings.
Ac-4,4-F2Pro-OMe. According to literature (17), based on comparison of coupling constant
patterns, and based on the assumption that the two counteracting stereoelectronic effects will
cancel each other out, the conformational behavior of Ac-4,4-F2Pro-OMe should be similar to
that of Ac-Pro-OMe. Fitting for Ac-4,4-F2Pro-OMe, always requires the use of 1H-19F
couplings, as the amount of vicinal 1H-1H couplings (two) is insufficient, even when assuming
only one conformer. Fitting one conformer yields results with high uncertainties and high
RMSD values, which, like Ac-Pro-OMe, indicate that one conformer is insufficient to explain
the data. This confirms the assumption that there will be an equilibrium between two or more
puckers. When fitting two conformers, not only do the puckers not correspond to Cγ endo or Cγ
exo (see Figure 3), but the results are also inconsistent between cis and trans forms and between
the two solvents. While it is clear that the possibility of other conformations than Cγ endo and
Cγ exo were never considered for this molecule, and that it was never analyzed so extensively
before, care has to be taken not to jump to conclusions too rapidly. When comparing the two
vicinal 1H-1H couplings of Ac-4,4-F2Pro-OMe with those of the other molecules, it is indeed
not unreasonable to assume that Ac-4,4-F2Pro-OMe is similar to Ac-Pro-OMe. Furthermore, it
is not clear to which extent the applied 1H-19F Karplus curves are accurate for fluoroprolines.
The results for Ac-4,4-F2Pro-OMe will indeed be the most sensitive to any inaccuracies in this
respect. Yet again, it can be stated that a proper validation of this relation is necessary. It can
thus be concluded that further research is needed in order to make a statement about the
puckering preferences of Ac-4,4-F2Pro-OMe.
xiii
Investigation of the cis/trans thermodynamics
The obtained Ktrans/cis values for all the samples can be found in Tables II and III. Three
trends can be observed. First, it can be seen that Ktrans/cis decreases with increasing temperature.
According to Le Châtelier’s principle, this means that the reaction is exothermic. This confirms
that the trans form is more stable than the cis form. Secondly, the observed Ktrans/cis values are
always the largest for Ac-(4R)-FPro-OMe, and the smallest for Ac-(4S)-FPro-OMe, with the
values for Ac-Pro-OMe and Ac-4,4-F2Pro-OMe falling in between. There is a correlation
between the puckering equilibrium and the cis/trans equilibrium by the n → π* interactions
(15) (vide supra). Since Ac-(4S)-FPro-OMe predominantly adopts the Cγ endo pucker, the
favorability of the cis conformer increases (also because of electrostatic repulsion between
fluorine and the carbonyl group), although the trans conformer is still more prevalent. For Ac-
(4R)-FPro-OMe, which predominantly adopts the Cγ exo pucker, the trans conformer is more
favored. Since the other two molecules are not biased towards one of the two puckers, their
Ktrans/cis values lie intermediate. Lastly, Ktrans/cis is always larger in D2O than in CDCl3. A
possible explanation was suggested by Siebler et al. (18), who also observed this trend. The
increased charge separation induced by n → π* interactions between adjacent carbonyl groups,
only occuring in the trans conformer, is better solvated in D2O than in CDCl3. TABLE II. Ktrans/cis values for the CDCl3 samples at different temperatures.
T(K) Ac-Pro-OMe Ac-(4R)-FPro-OMe Ac-(4S)-FPro-OMe Ac-4,4-F2Pro-OMe
283 4.36 4.12 1.80 3.29
288 4.22 4.48 1.80 3.18
293 4.03 4.32 1.74 3.10
298 3.89 4.20 1.70 3.03
303 3.76 4.00 1.68 2.95
308 3.67 3.89 1.65 2.92
TABLE III. Ktrans/cis values for the D2O samples at different temperatures.
T(K) Ac-Pro-OMe Ac-(4R)-FPro-OMe Ac-(4S)-FPro-OMe Ac-4,4-F2Pro-OMe
298 5.04 7.23 2.50 3.51
308 4.79 6.44 2.41 3.33
333 4.03 5.19 2.18 3.02
Investigation of the cis/trans kinetics
The obtained kex values for all samples can be found in Tables IV and V. Again, three trends
can be observed. First, one can see that kex increases with increasing temperature, because the
kinetic energy increases. Secondly, kex increases with increasing amount of fluorine atoms. This
is because the electron-rich fluorine atom draws electron density away from the amide bond,
decreasing its double bond character and therefore facilitating rotation around the bond (1, 7).
Finally, a clear solvent effect is visible. Not only is kex a lot lower in D2O than in CDCl3, but
also the values for Ac-(4R)-FPro-OMe and Ac-(4S)-FPro-OMe are clearly different in D2O,
while they are approximately the same in CDCl3. There is a change in charge separation in the
amide bond during rotation. This causes the carbonyl oxygen atom to have a larger partial
negative charge in a planar amide (equilibrium state) than in an orthogonal amide (transition
state). Therefore, the equilibrium state is more polar than the transition state. This explains why
rotation over the amide bond is slower in D2O, which shows better solvation for polar
compounds than CDCl3 and thus shows higher activation energy barriers to be overcome (19).
xiv
TABLE IV. kex values in s-1 for the CDCl3 samples at different temperatures. Determined via EXSY: all temperatures for Ac-
Pro-OMe and 303 K for the other samples. All others determined via selective inversion recovery.
T(K) Ac-Pro-OMe Ac-(4R)-FPro-OMe Ac-(4S)-FPro-OMe Ac-4,4-F2Pro-OMe
298 0.0973 ± 0.0073 0.2479 ± 0.0087 0.2058 ± 0.0205 0.7372 ± 0.0465
303 0.1849 ± 0.0181 0.4239 ± 0.0348 0.4327 ± 0.0105 1.4768 ± 0.0113
308 0.3295 ± 0.0389 0.7309 ± 0.0098 0.7544 ± 0.0125 2.5056 ± 0.0658
TABLE V. kex values in s-1 for the D2O samples at different temperatures. Determined via EXSY: all temperatures for Ac-Pro-
OMe and 308 K for the other samples. All others determined via selective inversion recovery.
T(K) Ac-Pro-OMe Ac-(4R)-FPro-OMe Ac-(4S)-FPro-OMe Ac-4,4-F2Pro-OMe
308 0.0354 ± 0.0080 0.1069 ± 0.0105 0.0415 ± 0.0063 0.1481 ± 0.0094
333 0.3693 ± 0.0086 0.8526 ± 0.0441 0.5200 ± 0.0269 1.7008 ± 0.0402
Summary
To summarize, PSYCHEDELIC has proven to be a very helpful tool for extracting 1H-1H and 1H-19F couplings. With the fitting methodology used, it was possible to confirm the
observations in literature concerning the puckering preferences of Ac-Pro-OMe, Ac-(4R)-FPro-
OMe and Ac-(4S)-FPro-OMe. Also thermodynamic and kinetic studies of the cis/trans
equilibrium could be confirmed and could be rationalized. However, there are still some
methodological challenges that need to be addressed. Handling strongly coupled protons in pure
shift spectra is a first important goal, as these are prevalent within (fluoro)prolines and limit the
number and accuracy of the measured couplings. A proper validation and refinement of the 1H-1H and certainly 1H-19F Karplus relations in the context of fluoroprolines is also needed in order
to comment on the conformational preferences of Ac-4,4-F2Pro-OMe, and in order to prove the
existence of a second puckering conformer for Ac-(4R)-FPro-OMe and Ac-(4S)-FPro-OMe. In
terms of thermodynamics and kinetics of the cis/trans equilibrium, extension of the temperature
dependent data points is necessary in order to determine precise values for ΔH°, ΔS°, and the
transition state activation-energy Ea, activation-enthalpy ΔH‡ and activation-entropy ΔS‡.
Acknowledgments
The 700 MHz is part of the interuniversitary NMR facility, jointly operated by UGent, UA and
VUB. The 500 MHz infrastructure is funded by the Hercules Foundation. Dr. Ilya Kuprov is
gratefully acknowledged for providing computationally optimized geometries for the
fluoroprolines, required as input in the 1H-19F Karplus relations.
References
1. M. Salwiczek, E. K. Nyakatura, U. I. Gerling, S. Ye and B. Koksch, Chem. Soc. Rev.,
41, 2135 (2012).
2. P. M. Hendrickx and J. C. Martins, Chem. Cent. J., 2, 20 (2008).
3. A. D. Bain, Progress in Nuclear Magnetic Resonance Spectroscopy, 43, 63 (2003).
4. E. S. Eberhardt, N. Panisik, Jr. and R. T. Raines, J. Am. Chem. Soc., 118, 12261 (1996).
5. D. Sinnaeve, M. Foroozandeh, M. Nilsson and G. A. Morris, Angew. Chem. Int. Ed.
Engl., 55, 1090 (2016).
6. N. G. Sharaf and A. M. Gronenborn, Methods Enzymol., 565, 67 (2015).
7. D. O'Hagan, Chem. Soc. Rev., 37, 308 (2008).
8. K. M. Thomas, D. Naduthambi, G. Tririya and N. J. Zondlo, Org. Lett., 7, 2397 (2005).
9. K. Zangger, Prog. Nucl. Magn. Reson. Spectrosc., 86-87, 1 (2015).
10. M. Foroozandeh, R. W. Adams, N. J. Meharry, D. Jeannerat, M. Nilsson and G. A.
Morris, Angew. Chem. Int. Ed. Engl., 53, 6990 (2014).
11. C. Altona and M. Sundaralingam, J. Am. Chem. Soc., 94, 8205 (1972).
xv
12. E. Diez, J. San-Fabian and J. Guilleme, Molecular Physics, 68, 49 (1989).
13. C. Thibaudeau, J. Plavec and J. Chattopadhyaya, J. Org. Chem., 63, 4967 (1998).
14. J. S. Alper and R. I. Gelb, J. Phys. Chem., 94, 4747 (1990).
15. L. E. Bretscher, C. L. Jenkins, K. M. Taylor, M. L. DeRider and R. T. Raines, J. Am.
Chem. Soc., 123, 777 (2001).
16. R. W. Newberry, B. VanVeller, I. A. Guzei and R. T. Raines, J. Am. Chem. Soc., 135,
7843 (2013).
17. C. Renner, S. Alefelder, J. H. Bae, N. Budisa, R. Huber and L. Moroder, Angew. Chem.
Int. Ed. Engl., 40, 923 (2001).
18. C. Siebler, B. Maryasin, M. Kuemin, R. S. Erdmann, C. Rigling, C. Grünenfelder, C.
Ochsenfeld and H. Wennemers, Chem. Sci., 6, 6725 (2015).
19. E. S. Eberhardt, S. N. Loh, A. P. Hinck and R. T. Raines, J. Am. Chem. Soc., 114, 5437
(1992).
1
1. Introduction
The title of this master thesis is ‘Effects of fluorination on the conformation of proline’.
In order to gain an understanding about the research that was conducted during this
thesis, an introduction to some concepts is necessary. First of all, the nature and
biological importance of the amino acid proline will be discussed. Secondly, some basic
concepts of fluorine chemistry will be introduced. Finally, a description of the use of
fluorine in protein research will be given, as well as some more specific details in using
fluoroprolines.
1.1. Proline
Proline is one of the 23 proteinogenic amino acids. Its most distinctive feature is that
its side chain binds with the backbone amide nitrogen (see Figure 1). This implies that
proline incorporates its Cα, its backbone amide N, and its aliphatic side-chain into a
five-membered ring, and that is has a secondary amine instead of a primary amine.
This makes proline a rather unique amino acid. Furthermore, within a peptide, this five-
membered cyclic side chain will impose steric constraints on the polypeptide backbone
conformation, making proline less compatible for incorporation in regular secondary
structures such as α-helices and β-sheets. Because proline lacks an amide hydrogen,
it cannot provide the hydrogen bond donor that is necessary for the formation of α-
helices and β-sheets. Often, proline is found in protein segments where unusual torsion
angles are required such as turns, loops or polyproline helices, vide infra [1].
Figure 1 Structure of L-Proline
There are two equilibria that determine the conformation of a proline residue within a
peptide: the endo/exo ring puckering equilibrium and the peptide bond cis/trans
equilibrium [2]. Ring pucker denotes the fact that the ring in cyclic compounds is usually
not flat, but adopts a certain conformation in which one or two of the ring atoms are
located out of the plane formed by the other atoms. In five-membered rings, when one
2
atom is located out of plane, this is called an envelope conformation, and when two
neighbouring atoms are located out of plane at opposite sides of the plane, this is called
a twist conformation. Generally, the ring pucker in proline is found to adopt either a Cγ
exo or a Cγ endo conformation, which are both envelope conformations where the Cγ
atom is out of plane. In case of the Cγ exo conformer, the Cγ atom and the C’ atom are
oriented away from each other while in case of the Cγ endo conformer, the Cγ atom
and the C’ atom are facing towards each other (see Figure 2) [3]. The cis/trans
equilibrium around the peptide bond is common for all amino acids and depends on
the ω dihedral angle (which is the angle defined by Cα-C’-N-Cα). The ω dihedral angle
is 0° for the cis conformation and 180° for the trans conformation (see Figure 3). In the
vast majority of cases, a peptide bond will adopt the trans conformation, mainly
because steric hindrance will occur between the two side-chains on the α-carbons
when the peptide bond adopts the cis conformation (see Figure 4). Again, proline
confirms that it is a unique amino acid, since the prolyl peptide bond is the only peptide
bond that will adopt the cis conformation in peptides at significant percentages. This is
because there will also be steric hindrance in the trans conformation, between the
prolyl side chain attached to the amide nitrogen and the side chain of the preceding
amino acid (see Figure 5), which causes the energy difference between both
conformers to become less. However, the trans conformation will still be the most
prevalent (Ktrans/cis of Ac-Pro-OMe = 4.6) [4].
Figure 2 L-proline. Left: Cγ exo conformation. Right: Cγ endo conformation. Based on a figure in [1]
Figure 3 Methyl acetyl-L-prolinate. Left: cis conformation. Right: trans conformation
3
Figure 4 Regular peptide bond. Left: trans. Right: cis. The dotted lines indicate the continuation of the peptide back bone.
Figure 5 Prolyl peptide bond. Left: trans. Right: cis. The dotted lines indicate the continuation of the peptide back bone.
The cis/trans equilibrium appears to be correlated with the ring puckering. Crystal
structure analysis of peptides and proteins revealed that trans-Xaa-Pro peptides are
mostly associated with the Cγ exo pucker, while a cis-Xaa-Pro peptide bond is most
often associated with a Cγ endo pucker [3]. It is assumed that this correlation is an
effect of the so-called n → π* interactions (vide infra) [5]. This suggests that if one
would be able to shift the puckering equilibrium towards one of the two conformers,
this could influence the cis/trans equilibrium. As will be discussed later, incorporation
of fluorine in proline residues can bias the ring puckering towards one of the two forms
[1].
In an n → π* interaction, the lone pair (n) electron orbital of one carbonyl oxygen
overlaps with the π* orbital of another carbonyl group (see Figure 6), providing an
electronically stabilizing effect. Increasingly, it is realized that this non-bonding
interaction could play a role in stabilizing proteins and peptides secondary structure [6,
7]. Because this interaction is a sub-van der Waals contact (the distance between the
atoms involved is less than 3.22 Å), a release of energy will take place due to
hyperconjugative delocalization. These interactions will polarize the electron density in
carbonyl groups, which could increase the strength of subsequent n → π* interactions
in which these are involved, thus leading to a cooperative effect. Because of an
increase of electron population of the π* orbital, the hybridization of the carbon atom
shifts from an sp²-like character to a more sp³-like character which effectively leads to
4
pyramidalization of the carbonyl group. This in turn causes the peptide bond to distort,
moving away from planarity. Such distortions will be relatively small though, usually
less than 1°.
Figure 6 Methyl acetyl-L-prolinate. n → π* interaction. Note that the amide bond adopts the trans conformation. Based on a figure in [7].
n → π* interactions can only take place when the peptide bond adopts the trans
conformation (as can be seen in Figure 6). This is the basis of the correlation of the
puckering equilibrium and the cis/trans equilibrium. An n → π* interaction will liberate
most stabilization energy when the angle of approach of the oxygen lone pair is closest
to the ideal Bürgi-Dunitz trajectory angle (τBD) of 109°, because these interactions can
also be seen as a nucleophilic attack of the oxygen lone pair on the carbonyl carbon.
The relative orientation of both is influenced by the φ torsion angle (defined by C’-N-
Cα-C’). As the ring pucker will determine the values for φ, it will in turn determine the
value of τBD (together with ψ). Therefore, it is clear the ring pucker will have an influence
on the cis/trans isomerism. In case of a Cγ exo pucker, φ will be around -60°, leading
to a τBD of approximately 100°, which is close to the ideal value of 109°, meaning that
an n → π* interaction will be more optimal (2.22 kJ/mol). This in turn, will contribute to
a higher trans/cis ratio. In case of a Cγ endo pucker, φ will not change that much (-
70°), due to the rigidity of the pyrrolidine ring, however, it will have a large influence on
τBD leading to a value of approximately 90°. This is a suboptimal value for the Bürgi-
Dunitz trajectory. This causes the stabilization energy of an n → π* interaction to be
rather low (0.59 kJ/mol), which will in turn decrease the trans/cis ratio [5, 7-9].
Next to α-helices and β-sheets, another important class of secondary structures are
the so-called polyproline I and II helices [10]. As the name suggests, these are
commonly found in peptides featuring repeated prolines, since these structures are not
stabilized by hydrogen bonds. Nevertheless, other residues than proline can adopt this
structural motif. The polyproline helix I (PPI) is a compact right-handed helix (φ = -75°
; ψ = 160° ; ω = 0° ; helical pitch = 5.6 Å/turn) in which all the peptide bonds adopt the
5
cis conformation. The polyproline helix II (PPII) is an extended left-handed helix (φ = -
75° ; ψ = 145° ; ω = 180° ; helical pitch = 9.3 Å/turn) in which all the peptide bonds
adopt the trans conformation (see Figure 7) [11]. Because of the correlated equilibria
mentioned above, proline derivatives that adopt the Cγ endo pucker have been found
to stabilize a PPI type helix, but destabilize a PPII type helix when incorporated in those
structures. Proline derivatives that adopt the Cγ exo pucker are found to stabilize a PPII
type helix (vide infra). As will be discussed later, incorporation of fluorine can influence
the ring pucker of proline residues [11].
Figure 7 Polyproline helices. Left: PPI helix, side view. Middle: PPII helix, side view. Top right: PPI helix, top view. Bottom right: PPII helix, top view. Figures taken from www.wikipedia.org.
In organic solvents, PPI helices are often encountered, while in aqueous solvents, PPII
helices are favoured [11]. This is because the carbonyl oxygen in a PPII helix is
perpendicular to the helix axis, making it more exposed to the solvent, while in a PPI
helix, the carbonyl oxygen is parallel to the helix axis and more shielded from the
solvent [12]. In case of a PPII helix, these exposed carbonyl oxygens can be stabilized
by hydrogen bonding in aqueous solvents but not or less in organic solvents. Since life
on earth is water based, PPII helices are therefore the most commonly encountered in
nature. Collagen and elastin, for example, which are two important fibrillar structural
proteins of special interest in the biomedical sciences due to their mechanical
properties, contain PPII helices [1]. PPII helices are also prevalent in proline rich
regions (PRRs), which play a role in protein-protein interactions, molecular recognition,
signal transduction and gene regulation, all typical processes to be targeted in rational
drug design [13].
6
Gaining an understanding of aforementioned processes, certainly when PRRs are
involved, is therefore of key importance. One thus needs to investigate the structure
and dynamics of the PRRs, which has been done by spectroscopic techniques such
as circular dichroism (CD), Förster resonance energy transfer (FRET) or Raman
optical activity [14-16]. However, all these techniques have the disadvantage that they
do not provide any residue-specific information. Nuclear magnetic resonance (NMR),
the technique par excellence to provide residue-specific information, is unfortunately
plagued by the absence of an amide proton in the proline residues, as well as the low
chemical shift dispersion encountered in disordered and proline-rich proteins. This
apparent incompatibility between proline residues and NMR leads us to consider the
incorporation of fluorine in the proline residues (vide infra). For this reason, it is
necessary to first discuss the structural effects that come with the introduction of
fluorine.
1.2. Organofluorine chemistry
Fluorine is the element with the highest electronegativity (χ=4 on the Pauling scale),
therefore the replacement of hydrogen (χ=2.1) with fluorine in organic molecules will
have important consequences. First of all, the C-F bond will not only be more polarized
than the C-H bond but will also be inversely polarized, i.e., the partial negative charge
will be on fluorine instead of carbon. This means that a C-F bond will have a more
electrostatic character than a C-H bond, which is mostly covalent. Additionally,
because of the high electronegativity of fluorine, the free electron pairs are quite tightly
bound to the fluorine atom, rendering the C-F bond less polarizable than one would
expect based on the electrostatic nature of the bond. For this reason, fluorine is
generally considered not to be a good hydrogen bond acceptor, however it will interact
with its environment via electrostatic-dipole interactions [17].
The C-F bond is the strongest known single bond in organic chemistry (bond
dissociation energy in kJ/mol: C-F = 441.0 ; C-H = 413.4 ; C-C = 347.7 [17]) and the
reason for this can again be found in the highly polarized nature of the bond. Fluorine
will have the tendency to attract the largest portion of the electron density, which means
that most of the electron density will be shifted to the fluorine atom. The high bond
strength can thus be attributed to the significant electrostatic interaction between Cδ+
and Fδ-. This electrostatic attraction will also lead to bond shortening when multiple
fluorine atoms are introduced on the same carbon atom, as the partial positive charge
7
of the carbon atom will increase. The polarization of the C-F bond not only has an
influence on bond strength and bond length but also on bond angles. As we go from
methane to fluoromethane, the H-C-H bond angle increases, because the C-F bond
pulls electron density away from the otherwise electron rich C-H bonds, effectively
relaxing electron repulsion between these bonds and causing them to spread out a
little [17]. Contrary to what one might expect based on steric hindrance and lone pair
repulsion, the F-C-H angle in fluoromethane is narrower than the H-C-H angle in
methane and this narrowing trend even continues in difluoromethane where the F-C-F
angle is the narrowest of all bond angles (H-C-H = 113.8° ; F-C-F = 108.4°). This trend
can be understood as a consequence of the fluorine atoms pulling the carbon’s p-
orbital electrons towards them, as the fluorine atom has a lower energy 2p orbital than
the carbon atom, giving more s-character to the carbon atom [17].
Another important effect introduced by fluorine is the so-called gauche effect. This
effect finds its origin in the highly polarized C-F bond which leads to a low energy
σ*C-F antibonding orbital, susceptible to hyperconjugation. Normally, one would initially
expect a 1,2-dihaloethane molecule to adopt the anti conformation due to the fact that
steric repulsion of the two halogen atoms is minimized when their respective torsion
angle is 180°. However, in the case of 1,2-difluoroethane, the lowest energy conformer
will be the gauche arrangement rather than the anti arrangement. The reason for this
is the stabilizing σC-H → σ*C-F hyperconjugative interactions (see Figure 8). In the anti
conformation, no such interactions can occur, while two such interactions can be
supported in the gauche conformation [17]. This is a characteristic feature of fluorinated
compounds, and cannot be observed in 1,2-dichloroethane and 1,2-dibromoethane.
Likely due to the σ*C-Cl/Br not being as low in energy, thus making orbital mixing less
efficient, and due to a larger steric repulsion between the more voluminous Cl/Br
atoms. Within 3,4-difluoroprolines (vide infra), the gauche effect will surely play an
important role in determining their conformation.
Figure 8 The stabilizing hyperconjugative interaction lies at the basis of the gauche effect. Based on a figure in [17].
8
From this point of view, one can understand that introducing fluorine in proline or
polyproline helices will have an influence on the conformation of the amino acid and
consequently on the global conformation of the peptide, since the gauche effect will
influence the ring pucker.
1.3. Applications of fluorine in protein sciences
Proline is certainly not the only amino acid where fluorination is of interest. In fact,
fluorination becomes more and more often exploited in protein sciences [18]. A short
overview of the possible applications of protein fluorination is provided in this section,
with particular attention to the use of fluoroprolines.
Fluorinated amino acids and fluorinated organic compounds in general are virtually
absent in nature, only five natural fluorine containing metabolites have been
discovered so far, only one of which is an amino acid, being 4-fluoro-L-threonine [19].
This implies that, in order to incorporate fluorinated amino acids in proteins, these have
to be synthesized. Forming carbon-fluorine bonds is challenging, typically involving
harsh reaction conditions (although specific classes of enzymes are known that can
facilitate the process) [18]. Despite the synthetic challenges, incorporating fluorinated
amino acids in proteins can be rewarding. Protein fluorination turns out to be a very
effective way to increase their stability, creating more resistance against unfolding by
heat, solvents, chemical denaturants or proteases. The reason is that protein folding
is mainly driven by a hydrophobic effect, and fluorination generally increases
hydrophobicity. Because of the stability of the carbon-fluorine bond (vide supra), the
fluorine functional group is chemically inert. Perfluorinated proteins are not only very
hydrophobic, but they also have interesting phase segregation behaviour, which is
known as the “fluorous effect”. Perfluorinated proteins will often not be soluble in both
aqueous and organic solvents, making them both hydrophobic and lipophobic.
Perfluorinating proteins can thus be a way to force proteins to interact, but whether this
is because of the “fluorous effect” or just because of the increased hydrophobicity, is
still an issue of debate among protein scientists. [18, 20]
Fluorine is often regarded as an isostere of hydrogen since they have comparable
atomic radii (1.2 Å for H and 1.47 Å for F). However, that claim is a bit too simplistic,
since the C-F bond is considerably longer than the C-H bond (1.35 Å versus 1.09 Å),
making fluorine more space-filling than hydrogen. Nonetheless, substituting fluorine for
9
hydrogen does not introduce much additional steric hindrance, which is confirmed by
the observation that fluorinated compounds mimic their non-fluorinated analogues
rather well. However, fluorination does introduce stereoelectronic effects (such as the
gauche effect) that alter conformational preference, as described above [18].
Fluorine is also a very useful element in terms of NMR, which is the central analytical
technique in this thesis. NMR is unrivalled when it comes to obtaining structural and
dynamic information of proteins and peptides. Usually 1H, 13C and 15N NMR techniques
are used to study proteins in solution (often with 13C or 15N isotopic labelling), but 19F
NMR is gaining more and more popularity. The 19F isotope is 100% naturally abundant
and is an NMR active nucleus (spin ½) with a gyromagnetic ratio nearly as high as 1H,
leading to a very good sensitivity (83% relative to 1H). Additionally, fluorine chemical
shifts span a very wide range of values (from -500 up to 500 ppm) and are extremely
sensitive to changes in local environment due to the large paramagnetic term in the
shielding of the 19F nucleus, making it an excellent probe for interaction studies. The
virtual absence of fluorine in biomolecules makes 19F an orthogonal probe, since
spectra can be recorded without background interference [21].
As stated before, proline is not the most NMR friendly amino acid. Certainly when
investigating PRRs, the lack of an amide proton is the most important drawback.
Indeed, the amide proton signal is the starting point for the assignment, which typically
involves TOCSY for peptides or 1H-15N HSQC for labelled proteins. A different issue is
that there is typically a lot of spectral overlap of the proline side-chain 1H resonances,
even within one residue. As can be understood from the previous paragraph,
introducing a limited number of fluorine atoms in one or multiple proline residues may
offer an opportunity in this respect. On the one hand, the 19F spectrum will be relatively
easy to interpret, since only a few signals will be visible. On the other hand, the 1H and
13C chemical shifts of the fluorinated proline will be significantly altered by the
fluorination, making it distinct from the other proline residues. A minor drawback is that
the 1H multiplets will generally display higher degrees of splitting due to the (often
large) scalar couplings with 19F. Fortunately, this can be overcome by the introduction
of 19F decoupling in the various NMR experiments. Thus, fluorination is a promising
strategy to obtain residue specific information about the proline residues. However,
one needs to take into account that fluorination will induce some conformational and
10
dynamical changes because of the stereoelectronic effects it imposes. These will be
discussed in the following paragraphs.
As mentioned before, the ring pucker of proline is in equilibrium between the Cγ exo
and Cγ endo forms. It has previously been reported that the ring puckering equilibrium
will be biased towards one of the two forms depending on the position of the fluorine
and the stereochemistry. The ring puckering equilibrium in (4R)-fluoro-L-proline (Flp)
will be biased towards the Cγ exo form because of the gauche effect, which induces a
preference for the fluorine substituent to adopt a gauche orientation relative to the
amine (see Figure 9). The opposite is true in the case of (4S)-fluoro-L-proline (flp),
which will bias the ring puckering equilibrium towards the Cγ endo form for the same
reason as in the case of (4R)-fluoro-L-proline (see Figure 9). Since the proline ring
pucker is correlated with the stability of the cis and trans peptide bond, and thus also
the stability of polyproline helices (vide supra), introducing a (4R)-fluoro-L-proline in a
PP-II type helix can be expected to stabilize that helix, while a (4S)-fluoro-L-proline
destabilizes it [1]. Horng et al. [11] studied the behaviour of uniform decafluoroprolines
– (Pro)10, (Flp)10 and (flp)10 – in aqueous solutions by means of temperature dependent
CD spectroscopy. The first thing that could be noticed, was the difference between the
spectra of (Pro)10 and (Flp)10 on the one hand and the spectra of (flp)10 on the other
hand. The former two showed spectra that are a fingerprint of a PPII helix, while for
the latter, the spectrum was that of a mixture of PPI and PPII helices. This already
confirms that flp destabilizes the PPII conformation. To assess if Flp stabilizes the PPII
conformation, the estimated Tm values for (Pro)10 and (Flp)10 were compared, which
were 27°C and 53°C respectively. An increased melting temperature indeed indicates
an increased stability. (Flp and flp nomenclature was taken over from [11])
Figure 9 Newman projections of fluoroprolines. Left: (4R)-fluoro-L-proline. Right: (4S)-fluoro-L-proline. When placing the F substituent gauche relative to the amine substituent, the Cγ exo and Cγ endo conformation can be found respectively.
11
Furthermore, fluorination will have an influence on the extent with which n → π*
interactions occur in folded proteins. Since n → π* interactions are only possible when
the amide bond is in the trans conformation, introducing a fluoroproline that favours
the Cγ exo pucker will preorganize the backbone in such a way that favours n → π*
interactions (vide supra) [1, 6, 7].
Lastly, fluorination does not only influence Ktrans/cis, partly due to its influence on the
ring pucker, and partly due to the electrostatic repulsion of the fluorine atom and the
carbonyl oxygen, but it will also influence the kinetics of the cis/trans isomerism. The
electron withdrawing effects of the fluorine atom will cause the C-N bond in the amide
group to lose part of its double bond character, making rotation around this bond easier
and thus accelerating the cis/trans isomerism [1].
So far, synthesis of the following fluoroprolines has been reported: (4R)-fluoro-L-
proline, (4S)-fluoro-L-proline, 4,4-difluoro-L-proline [4, 5, 8, 22], (3R)-fluoro-L-proline,
(3S)-fluoro-L-proline [23, 24], 3,3-difluoro-L-proline [25], (3R,4R)-difluoro-L-proline [26]
and (3R,4S)-difluoro-L-proline [27] (see Figures 10, 11 and 12). Except for 3,3-F2Pro
and both 3,4-F2Pro’s, conformational analyses have been conducted by means of
NMR and crystallography, the results of which can be seen in Table 1.
In this research project, the studied (fluoro)prolines were derivatized as such: the
carboxylic acid function was converted into a methyl ester and the amine was
acetylized (Ac-(F)Pro-OMe derivatives). The following molecules will be investigated
here: non-fluorinated (Ac-Pro-OMe), monofluorinated at the 4 position (Ac-(4R)-FPro-
OMe and Ac-(4S)-FPro-OMe), and difluorinated at the 4 position (Ac-4,4-F2Pro-OMe).
These molecules were all already excessively studied in the past in terms of puckering
and cis/trans. The goal of this thesis is to reinvestigate the ring pucker by measuring
and analysing the 1H-1H and 1H-19F scalar couplings using advanced NMR
methodology, that enables the extraction of individual coupling constants from crowded
spectra. Also a systematic study of the cis/trans conformational equilibrium and its
kinetics will be performed. The methodology that is explored in this thesis – as
discussed in the following chapters – will be needed for future structural and dynamic
studies of the much less well-studied 3-fluoroprolines, 3,3-difluoroproline and the 3,4-
difluoroprolines.
12
Figure 10 The 4-fluoro-L-prolines. Left: Ac-(4R)-FPro-OMe. Middle: Ac-(4S)-FPro-OMe. Right: Ac-4,4-F2Pro-OMe.
Figure 11 The 3-fluoro-L-prolines. Left: Ac-(3R)-FPro-OMe. Middle: Ac-(3S)-FPro-OMe. Right: Ac-3,3-F2Pro-OMe.
Figure 12 The 3,4-difluoro-L-prolines. Left: Ac-(3R,4R)-F2Pro-OMe. Right: Ac-(3R,4S)-F2Pro-OMe.
Table 1 Literature values for conformational preferences of the Pro pyrrolidine ring in Ac-Pro-OMe derivatives [1].
Amino Acid Predominant Pucker Ktrans/cis (298K)
Pro Cγ endo 4.6
(4R)-FPro Cγ exo 6.7
(4S)-FPro Cγ endo 2.5
4,4-F2Pro Cγ endo 3.4
(3R)-FPro Cγ exo 8.9
(3S)-FPro Cγ endo 4.3
13
2. Experimental Methods
In this chapter, methodologies for the assignment of the spectra, measuring coupling
constants and performing kinetic studies will be discussed.
2.1. Introductory information
2.1.1. The scalar coupling
The basics of NMR can be found in the supporting information (section 6.1.1) and will
not be discussed here. However, since the concept of scalar couplings between spins
will be central in this thesis, a brief description will be provided here. A scalar coupling
is a through-bond interaction between nuclei, in which the nuclear spin of one nucleus
affects the energy levels of the other. The mechanism of the coupling involves the
electrons within the chemical bonds between the nuclei. The nuclear spin of one
nucleus couples to the electron spin of the bonding electrons, which, in turn, are
coupled to the nuclear spin of the second nucleus. This will lead to a decrease or
increase of the nuclear spin’s energy level depending on the spin-state (up or down)
of the neighbouring nuclear spin. This causes the peaks of the involved spins – the
effect is mutual – to split in two, with a splitting equal to the scalar coupling constant J
(in Hz). The value of J is field independent. It is determined by the type of interacting
nuclei, the number and type of bonds that separate them, as well as the molecular
geometry (bond distances, bond angles, torsion angles). The latter makes the scalar
coupling a very powerful tool for conformational analysis (see Chapter 3). A scalar
coupling will be denoted as nJAX, in which n is the number of separating bonds, and A
and X are the types of interacting nuclei (e.g. 3JHH).
2.1.2. Studied samples
The following molecules were studied: Ac-Pro-OMe, Ac-(4R)-FPro-OMe, Ac-(4S)-
FPro-OMe and Ac-4,4-F2Pro-OMe (see Figure 13), dissolved in either CDCl3 or D2O.
Figure 13 From left to right: Ac-Pro-OMe, Ac-(4R)-FPro-OMe, Ac-(4S)-FPro-OMe and Ac-4,4-F2Pro-OMe.
14
2.2. Assignment of the spectra
The chemical shift assignment of each compound was carried out for three nuclei,
being 1H, 13C and 19F. For 1H and 19F, 1D spectra were recorded while for 13C, the
assignment was based entirely on 2D 1H-13C correlation spectra. The following 2D
NMR techniques were used in the assignment of the spectra: 2D 1H-1H COSY, 2D 1H-
13C HSQC, 2D 1H-13C HMBC, 2D 1H-1H NOESY and 2D 1H-19F HOESY. A short
description of how these different kinds of spectra need to be interpreted can be found
in the supporting information (section 6.1). Before discussing the chemical shift
assignment strategy, some general features of the spectra will be discussed. This will
be necessary to understand some parts of the assignment process (vide infra).
2.2.1. General features of the spectra
First of all, as described in the previous
chapter, two equilibria will play a role in
the investigated samples. Because the
nitrogen is acetylated, the resulting
tertiary amide group can undergo
cis/trans isomerism, similar as in
peptides. The interconversion between
the cis conformer and the trans
conformer is slow on the NMR frequency
timescale. This means that for each
proton in the molecule, two resonances can be seen, one from the molecule that
adopts the trans conformation, and another where the molecule adopts the cis
conformation (see Figure 14). The most intense resonance will be referred to as the
major resonance, and the other as the minor. Based on literature we can expect the
major and the minor to correspond with the trans and the cis conformers respectively.
The ratio of the integrals of both peaks will reflect Ktrans/cis (see chapter 4).
Figure 14 The α 1H resonances of Ac-(4S)-FPro-OMe in CDCl3. 298 K. 500 MHz.
15
Secondly, the hydrogen atoms in
the ring are diastereotopic,
implying that two 1H’s on the same
carbon atom, generally do not
have the same chemical shift value
(see Figure 15). For this reason,
some effort has to be invested into
distinguishing the pro-R and pro-S
resonances of the β and δ 1H’s.
The same is true for the γ 1H’s in
the case of Ac-Pro-OMe, and the γ 19F’s in the case of Ac-4,4-F2Pro-OMe.
Lastly, except for Ac-Pro-OMe,
where no fluorine atom is present,
the resonances of all 1H’s in the
vicinity of the fluorine atom will
show peak splitting due to a scalar
coupling with the 19F nucleus. This
scalar coupling is most striking for
the γ 1H of the monofluorinated
analogues, which show a 2JHF
splitting of ca. 50 Hz, as is evident
from Figure 16.
These combined effects can make the spectrum rather crowded in some places,
leading to a lot of spectral overlap, certainly in the β- and the δ-region (see Figure 17).
2.2.2. General assignment strategy of the 1H spectra
The methyl signals can easily be discerned in the 1H spectra as four large singlet peaks
(see Figure 18). These correspond to the two different methyl groups, both showing a
major and a minor form due to cis/trans isomerism (vide supra). The signals are intense
since they should account for three protons, and because they are singlets. These
methyl signals are always found around 2 and 3.75 ppm. The former correspond to the
acetyl methyl, while the latter correspond to the methoxy methyl. This assignment is in
agreement with the chemical shift values, since 1H signals of a methyl adjacent to a
Figure 15 The β 1H resonances of Ac-(4S)-FPro-OMe in CDCl3. 298 K. 500 MHz.
Figure 16 The γ 1H resonances of Ac-(4S)-FPro-OMe in CDCl3. 298 K. 500 MHz.
16
carbonyl typically exhibit chemical shifts between 1.90 and 2.25 ppm, while the
chemical shifts of a methyl neighbouring an oxygen atom range from 3.5 to 4 ppm. The
NOESY spectrum also provides confirmation of this assignment (vide infra).
Figure 17 The β and γ region of the 1H spectrum of Ac-Pro-OMe in D2O. 298 K. 500 MHz. Top: 1D 1H. Bottom: 1D PSYCHE. Note that the trans γ protons cannot be distinguished. There are also artefacts due to strong coupling between the trans γ
and trans pro-R β protons at around 1.94 ppm.
Figure 18 1D 1H spectrum of Ac-(4S)-FPro-OMe in CDCl3. 298 K. 500 MHz.
17
The resonances of the α proton are also easily identified. For proline, these resonances
are expected around 4.4 ppm [28], though for the fluorinated analogues, the
resonances are a bit shifted to higher chemical shift values due to the presence of
fluorine (see Figure 18). They have the appearance of doublets of doublets, or triplets
or doublets, depending on the value of the coupling constants (see next section). In
the 1H-13C HSQC, these signals give positive (CH multiplicity) correlations, which is an
additional confirmation that these are the α 1H resonances.
Figure 19 2D 1H-1H COSY spectrum of Ac-(4S)-FPro-OMe in CDCl3. 298 K. 700 MHz. Start of sequential walk, correlations between α and β.
The β, γ and δ 1H resonances can be assigned by making use of the COSY spectrum
and a technique called sequential walking (see Figure 19). The α resonance only
shows correlations to the β resonances (both pro-R and pro-S) in the COSY. The major
α resonance will only show correlations to the major β resonances, and the minor α
resonance only to the minor β resonances, as the major and the minor are two different
spin systems. This reasoning will be valid for all correlations and will no longer be
explicitly mentioned from this point on. Next to a correlation with the α resonance, the
18
β resonances should also show a correlation to the γ resonance(s), and the γ
resonance(s) should show a correlation to the δ resonances. This assignment is further
validated by the 1H-13C HSQC. The resonances assigned to the β and δ 1H’s (and the
γ 1H’s for the Ac-Pro-OMe samples), should give negative (CH2 multiplicity)
correlations in the HSQC, and both protons within a CH2 fragment should correlate to
the same 13C chemical shift. In the case of the monofluorinated samples, the γ
correlations will be positive, the chemical shift should be around 5 ppm, and a large
2JHF coupling should be visible (see Figure 20).
In a last step, the stereospecific assignment (i.e., pro-R and pro-S) needs to be
performed for the β and δ resonances (and γ in case of the Ac-Pro-OMe samples).
This can be done with help of the NOESY spectrum. For the monofluorinated samples,
the strategy is rather straightforward. The β proton that shows the most intense nOe
cross-peak with the α signal is assumed to be on the same side of the five-membered
ring as the α proton. This is also verified by looking at the nOe cross-peaks with the γ
resonance, where a similar reasoning holds (see Figure 20). For the δ protons, one
can check the intensities of the nOe cross-peaks to the γ signal and to the already
assigned β signals. In all cases, it is again assumed that the most intense cross-peaks
indicate that the corresponding pair of protons are on the same side of the ring.
Figure 20 2D 1H-1H NOESY spectrum of Ac-(4R)-FPro-OMe in CDCl3. 298 K. 500 MHz. Most intense nOe cross peak indicates that this β hydrogen is on the same side of the ring as the γ hydrogen.
19
The NOESY also offers a possibility to confirm whether the major resonances are
indeed originating from the trans conformer. The methyl resonance of the acetyl group
of the major set of resonances shows an nOe cross-peak with the δ resonances, in
agreement with what would be expected when the molecule adopts the trans
conformation. For the minor set of resonances, no such cross-peak is observed, in
accordance with the cis conformation.
2.2.2.1. Some special cases
2.2.2.1.1. The Ac-4,4-F2Pro-OMe samples
The biggest nuisance in the spectra of Ac-4,4-F2Pro-OMe is the fact that the large
couplings with both 19F’s make the β (2.5–3 ppm) and δ (around 4 ppm) regions rather
unintelligible (see Figure 21). For this reason, 19F decoupled spectra were recorded
which greatly aided the assignment for these samples. Furthermore, the diastereotopic
assignment of the δ hydrogens turned out to be unattainable based on the 1H-1H
NOESY alone. Therefore, a 2D 1H-19F HOESY was recorded, in which the same
strategy for distinguishing pro-R and pro-S as in the NOESY (vide supra) was used.
Figure 21 Ac-4,4-F2ProOMe in CDCl3, 298K, 400MHz. Bottom: 1D 1H. Top: 1D 1H with 19F decoupling.
2.2.2.1.2. Overlap and strong coupling
In this part, the most difficult cases of spectral overlap will be discussed. This will be
important for the next part, as this overlap may cause trouble in measuring couplings.
For Ac-Pro-OMe in CDCl3, all four β and γ protons possess similar chemical shifts,
leading to a very crowded region. At about 2.20 ppm, the pro-R β signal of the cis
20
conformer and the pro-S β signal of the trans conformer are found, with their multiplets
overlapping. This complicates multiplet analysis in a later stage. At about 1.95 ppm,
the signals of both the pro-R and pro-S γ 1H of the cis conformer possess nearly the
same chemical shift. Since these protons are coupled, this is a case of so-called strong
coupling, i.e., whereby the difference in resonance frequency is of the same order or
smaller that the scalar coupling. This introduces second-order effects on the multiplet
structures, severely complicating multiplet analysis, or, as is the case for these γ 1H’s,
making the protons spectroscopically indistinguishable. At about 2.02 ppm, the
multiplets assigned to the pro-R β and pro-R γ protons of the trans conformer are found
– and show some degree of strong coupling – as well as the acetyl methyl signal of the
cis conformer.
Also in D2O, spectral overlap and strong coupling complicate matters (see Figure 17).
At 1.94 ppm, signals from four protons overlap, being pro-R β, pro-R γ and pro-S γ of
the trans conformer and one of the methyl’s of the cis conformer. The two trans γ
protons and the pro-R β proton are thus strongly coupled. Also both δ protons of the
trans conformer have close chemical shifts.
The 1H spectra of Ac-(4R)-FPro-OMe do not pose many challenges. The only case of
complications by strong coupling can be found in CDCl3, where the multiplets of both
the δ hydrogens of the trans conformer are overlapping.
Figure 22 The β 1H region of Ac-(4S)-FPro-OMe in D2O. 298 K. 500 MHz. The two large signals are coming from the trans conformer. Top: 1D 1H. Bottom: 1D PSYCHE. Note the strong coupling artefacts (circled) in the PSYCHE.
21
For Ac-(4S)-FPro-OMe, there is a big contrast between the two solvents in terms of
overlap. In CDCl3, there is overlap between the δ protons of the trans and cis
conformers, but fortunately there are no strong coupling complications, since the pro-
R and the pro-S within one conformer are always resolved. However, in D2O, the β
protons of the trans conformer are very strongly coupled and nearly indistinguishable,
even in the pure shift spectrum (vide infra). (see figure 22).
For Ac-4,4-F2Pro-OMe
in D2O, the pro-R and
pro-S δ signals appear
to have practically the
same chemical shift
(see Figure 23). In
CDCl3, there are no
significant overlap
problems.
Overlap can be most
often resolved by using
pure shift experiments
[29] like PSYCHE [30] and PSYCHEDELIC [31] (vide infra). Strong coupling, however,
remains an issue in these experiments, as is clear from Figures 17, 22 and 29. This
will be elaborated in more detail in a later section (2.3.1.2).
The 1H chemical shift values of all samples can be found in the supporting information
(section 6.2).
2.2.3. General assignment strategy of the 13C spectra
Starting from the 1H assignment, one can use the 2D 1H-13C HSQC to obtain the 13C
chemical shift values of all carbons, except the quaternary ones (see Figure 24). To
assign these quaternary carbons, the 2D 1H-13C HMBC was used. The 1H signals of
both the acetyl group and the methoxy group will show correlations to 13C resonances
at about 170 ppm, which is the expected position for carbonyl carbons of ester and
amide groups. These correlations thus complete the 13C chemical shift values of all
compounds, which can be found in the supporting information (section 6.3).
Figure 23 19F decoupled 1D 1H spectrum of Ac-4,4-F2Pro-OMe in D2O. 298 K. 400 MHz. Note the very steep roof effect of the δ of the trans conformer.
22
Figure 24 2D 1H-13C HSQC of Ac-(4S)-FPro-OMe in CDCl3. 298K. 700 MHz.
2.2.4. General assignment strategy of the 19F spectra
The assignment of the 19F spectra of the monofluorinated samples is straightforward,
only two peaks are found, a major and a minor one, which correspond to the trans and
cis conformer respectively. The use of 1H decoupled spectra was a prerequisite in
some cases, as the many 1H-19F couplings led to significant overlap between both
multiplets (see Figure 25). In case of the difluorinated samples, four signals – two minor
(cis) and two major (trans) doublets – are observed in the 1H decoupled spectra. The
doublets result from the 2JFF coupling and show distorted intensities (the “roof” effect)
due to the onset of strong coupling conditions. In this case, a stereospecific assignment
of both 19F signals is required, which can be achieved using the 2D 1H-19F HOESY
spectrum, as described in section 2.2.2. 19F chemical shift values can be found in the
supporting information (section 6.4).
2.3. Measuring coupling constants
The largest portion of experimental work was devoted to measuring coupling
constants, as these are essential for the puckering analysis (see Chapter 3). The way
these coupling constants were measured will be discussed in this section, as well as
the complications that arose. The theory behind scalar couplings was already
discussed in section 2.1.
23
Figure 25 Left: 1D 19F spectrum without (bottom) and with (top) 1H decoupling of Ac-(4S)-FPro-OMe in D2O. Right: 1D 19F spectrum with (bottom) and without (top) 1H decoupling of Ac-4,4-F2Pro-OMe in D2O. Both recorded at 298 K and on a 500
and 400 MHz spectrometer respectively.
2.3.1. Measuring 1H-1H couplings
2.3.1.1. PSYCHEDELIC
To measure 1H-1H couplings, the recently developed PSYCHEDELIC experiment was
used [31]. PSYCHEDELIC stands for Pure Shift Yielded by CHirp Excitation to DELiver
Individual Couplings. In this experiment, which is based on the PSYCHE pure shift
experiment [30], a 1H is selected, and a 2D J-resolved spectrum is generated that
contains only the couplings to this 1H as simple doublets (see Figure 26). In the 2D J
spectrum, the doublets are dispersed along a –45° pattern. This can be tilted by 45° to
align the splitting fully parallel along the F1 dimension, leaving only chemical shift
information along F2, and providing a resolution similar to the PSYCHE pure shift
experiment. In principle, one can excite multiple 1H’s at the same time, but to retain the
simple doublet splittings, and thus unambiguous interpretation, these 1H’s may not
have mutual coupling partners amongst the 1H’s to which the couplings are measured.
For example, one can always simultaneously select the same 1H of both the cis and
trans conformer safely, since these belong to different spin systems (see Figure 26).
In order to actually measure the value of the scalar coupling, a slice along F1 was
extracted from the spectrum for each doublet and smoothed in order to more accurately
measure the distance between the two peaks than in the 2D spectrum. This smoothing
consists of increasing the digital resolution of the extracted slice by applying an inverse
Fourier transform, back to the time domain, application of additional zero-filling, and
finally Fourier transformation of the result to generate the smoothed trace. Further
24
Lorentz-to-Gauss resolution enhancement could also be applied prior to this step (see
Figure 27). Now, a more accurate estimate of the peak splitting and thus the coupling
constant can be made.
The values of all measured coupling constants can be found in the supporting
information (section 6.5).
Figure 26 PSYCHEDELIC of Ac-(4R)-FPro-OMe in CDCl3. 298 K. 500 MHz. pro R β (major + minor) excitation.
Figure 27 Left: Extracted 1D slice of Figure 15. Right: Same 1D slice after smoothing. There is an intensity decrease due to a Lorentz-to-Gauss window function that was additionally applied.
25
2.3.1.2. Encountered difficulties
A first difficulty that one can encounter is that the coupling is so small that the peak
splitting observed along F1 is close to or smaller than the linewidth of an individual
signal (which here is typically about 1 Hz). In that case, the two peaks of the doublet
cannot be resolved, as in Figure 27, and possibly only one peak maximum is seen,
such as in Figure 28. Sometimes this can be overcome by applying further Lorentz-to-
Gauss resolution enhancement. The latter not only transforms the Lorentzian
lineshape to the more favourable Gaussian one, it can also trade sensitivity for
resolution (see Figure 28). However, sometimes even this did not help. In such cases,
the only information that could be reported, was an upper limit for the coupling of 1 Hz.
Figure 28 Ac-(4R)-FPro-OMe in CDCl3. 298 K. 500 MHz. γ excitation, pro S δ observation. Left: line broadening parameter set to -1.5. Right: line broadening parameter set to -5.
Another problem that was encountered is spectral overlap, which was already
discussed in section 2.2.2.1.2. When there is overlap between multiplets, the selective
radio-frequency pulse used to select protons in PSYCHEDELIC cannot distinguish
between both protons, selecting them both. As mentioned before, this is only an issue
when these two protons have a mutual coupling partner, as then no longer simple
doublets will be obtained along F1.
A final, and more serious, problem arises when protons are very strongly coupled, i.e.,
the coupling between them is of the same order of magnitude as their resonance
frequency difference. When a proton is selected in PSYCHEDELIC that couples to both
of these protons, strong coupling artefact peaks arise at the position of the strongly
coupled spins, impeding straightforward spectral interpretation and coupling
measurement (see Figure 29). If the strongly coupled protons possess practically the
same chemical shift, a peak splitting may be obtained that is assumed to be close to
26
the average of the two couplings. If, on the other hand, the two strongly coupled spins
are selected in PSYCHEDELIC, the sum of their couplings to a third spin can be
measured on their mutual coupling partner. There were four such very problematic
cases. In three of them, either a sum or average could still be measured. These cases
were: the γ protons of the cis conformer of Ac-Pro-OMe in CDCl3, the δ protons of the
trans conformer of Ac-(4R)-FPro-OMe in CDCl3, and the β protons of the trans
conformer of Ac-(4S)-FPro-OMe in D2O. The case were no couplings could be
measured was the γ protons of the trans conformer of Ac-Pro-OMe in D2O, as these
two very strongly coupled protons were additionally overlapping (and thus strongly
coupled) with one of the two β protons. In general, accurately measuring couplings
between and with strongly coupled spins still remains a methodological challenge.
Figure 29 Ac-(4S)-FPro-OMe in D2O. 298 K. 500 MHz. γ excitation, β region. a: pro S β cis. b: pro R β cis. Strong coupling artefacts can be observed (circled) for the trans conformer.
2.3.2. Measuring 1H-19F couplings
For the monofluorinated molecules, 1H-19F couplings can simply be measured from a
1D PSYCHE pure shift experiment [32]. In a pure shift experiment, 1H-1H splittings
have been eliminated, meaning only the 1H-19F couplings, will be present as doublets
and thus easily measurable (see Figure 30). For Ac-4,4-F2Pro-OMe, there are two
fluorines present, meaning that in the 1D PSYCHE, the peaks of 1H’s that couple with
19F will be split into a doublet of doublets. This is problematic, because even if one can
then accurately extract these couplings from the spectrum, one in principle cannot
27
know which of the fluorine atoms is involved in the particular coupling. An alternative
experiment is a heteronuclear 1H-19F PSYCHEDELIC experiment developed in our
group (unpublished result). The experiment works similar as a homonuclear
PSYCHEDELIC experiment, except that now a single fluorine is selected, so that 1H-
19F splittings will be observed along F1, while all 1H-1H and 1H-19F splittings are
suppressed along F2.
All 1H-19F coupling values can be found in the supporting information (section 6.6).
Figure 30 1D PSYCHE Ac-(4R)-FPro-OMe in CDCl3. 298 K. 500 MHz.
1H-19F couplings are generally larger than 1H-1H couplings, so there were never any
issues with them being smaller than the linewidth. However, spectral overlap and
strong coupling between protons can pose a problem. Sometimes it was not possible
to measure a 1H-19F coupling from the 1D PSYCHE due to overlap. For example, for
Ac-(4S)-FPro-OMe in CDCl3, the pro-R δ doublets of both the cis and trans conformer
overlap and also the methoxy peaks are interfering. Fortunately, in this case, it was
possible to measure the 1H-19F couplings from the 2D 1H-13C HSQC, as these signals
are resolved along the 13C dimension, although this is somewhat less accurate. In
some cases where couplings could not be measured in the 1D PSYCHE, the HSQC
could also not provide a solution. For the trans conformer of Ac-4,4-F2Pro-OMe in D2O,
the two δ 1H’s are very strongly coupled. Similarly as described above for homonuclear
PSYCHEDELIC, it was in this case only possible to measure an average of the
couplings.
28
2.4. NMR techniques for measuring relaxation and kinetics
In this section, a short description will be provided of the techniques used in Chapter 4
for the study of the cis/trans exchange process.
2.4.1. Inversion recovery
When the z-magnetization of a spin is perturbed from its equilibrium value (𝐼𝑧0) after an
rf-pulse, it will relax back to its equilibrium state with a unique longitudinal relaxation
time constant T1. The inversion recovery experiment is used to measure the value of
T1 for each resonance in the spectrum. After a p pulse, the z-magnetization undergoing
T1 relaxation as a function of time can be described by the following equation.
𝐼𝑧(𝑡) = (1 − cos(𝑝))𝐼𝑧0𝑒−𝑡𝑇1
In the inversion recovery experiment, a 180° pulse is applied, which will invert the z-
magnetization (cos(p) = -1), which will then start to relax back to its equilibrium state
according to the above equation. After a certain delay, a 90° pulse is applied and
acquisition is performed. The intensity of the observed signal will depend on the
duration of the delay in the same way as the above equation. By doing a series of such
measurements with a variable delay, one can plot the obtained intensities as of function
of the delay, and fit the above function to the data points. In that way, one can retrieve
the value for T1.
2.4.2. Selective inversion recovery
When two spins are in chemical exchange with one another, the selective perturbation
of the z-magnetization of one of these spins will, in turn, influence the z-magnetization
of the other. The extent of this influence will depend on the exchange rate kex, which is
the sum of the forward (k1) and the backward (k-1) rate constants of the exchange
reaction. If one would selectively invert the z-magnetization of one of the two spins,
and track the intensity change of the other spin as a function of time as in a regular
inversion recovery experiment, a value for kex can be assessed. Such inversion can be
achieved by either applying a selective soft 180° pulse on one of the two exchanging
spins, or by applying a 90°x – delay – 90°-x pulse sequence, in which the delay is the
reciprocal value of two times the frequency difference of both resonances. In the latter
experiment, the original orientation of the on-resonance peak’s z-magnetization will be
29
preserved after this sequence, while the off-resonance peak’s z-magnetization will
have been inverted.
2.4.3. 2D EXSY
Another method for measuring kex is a 2D 1H-1H EXSY (EXchange SpectroscopY). In
terms of pulse sequence, an EXSY is identical to a NOESY, but the cross-peaks of
interest are in this case not caused by the nOe, but by chemical or conformational
exchange between the two spins. These peaks will always possess the same sign as
the diagonal peaks. The value for kex can be retrieved from the diagonal and cross-
peak peak volumes after a certain EXSY mixing time.
A more theoretical explanation concerning selective inversion recovery and EXSY will
be provided in Chapter 4, as well as the reason why different kinds of experiments
were used for the same purpose.
30
3. Puckering analysis
In this chapter, information about the puckering equilibrium of the studied molecules is
derived from the measured couplings. First, the mathematical description of five-
membered ring conformations will be addressed. Next, the relation between scalar
couplings and torsion angles will be discussed. After that, data fitting procedures will
be explained. Lastly, the results and conclusions for the studied molecules will be
presented.
3.1. Mathematical description of five-ring conformations
The phenomenon of ring pucker – the deviation of planarity – was already discussed
in Chapter 1. The question is now, how to conveniently describe the ring pucker.
Annotations like Cγ endo and Cγ exo are very limited given the amount of conformations
that can potentially be described, and are also only relevant for proline related systems.
It is clear that a more general way of describing conformations is needed. In general,
in order to describe an N-membered ring conformation, N-3 parameters are needed,
this means that two parameters will be needed for five-membered rings [33]. There are
two formalisms that propose such parameters. The Cremer-Pople formalism, which is
based on Cartesian coordinates [34], and the Altona-Sundaralingam formalism, which
is based on endocyclic torsion angles [35]. Since it is more convenient to relate torsion
angles to scalar couplings in a later step, the Altona-Sundaralingam formalism will here
be preferred over the Cremer-Pople formalism [36].
The Altona-Sundaralingam formalism reduces the description of the ring pucker to two
puckering parameters P and νmax that are correlated to the endocyclic torsion angles
νi (i = 0 – 4) in the following manner:
𝜈𝑖 = 𝜈𝑚𝑎𝑥 cos (𝑃 +4𝜋
5𝑖)
This is in fact a set of five equations, one for each endocyclic torsion angle. The inverse
relationships to obtain P and νmax from the torsion angles are:
tan(𝑃) = −𝜈1 + 𝜈2 − 𝜈3 + 𝜈4
2𝜈0 (sin (𝜋5) + sin (
2𝜋5))
31
𝜈𝑚𝑎𝑥 =𝜈0
cos (𝑃)
As is evident from the above formula, P will be dependent on the numbering scheme.
Throughout this thesis, the numbering scheme as depicted in Figure 31, was used [36].
Figure 31 Used numbering scheme for the endocyclic torsion angles.
The two puckering parameters proposed by Altona and Sundaralingam can be
interpreted in the following manner. P denotes the pseudorotation phase angle (0° –
360°), and indicates which atom(s) is (are) out of plane. When P is an odd multiple of
18°, the conformation will be a true envelope, while even multiples of 18° correspond
to pure twist conformations. Of special interest for this investigation are the P values
for the Cγ exo and Cγ endo conformations, which are 18° and 198° respectively for the
numbering scheme used here. νmax denotes the puckering amplitude (≥ 0°), and
indicates how much the atom(s) is (are) tilted out of plane. A puckering amplitude of 0°
would thus indicate that the ring is flat [36].
Because of the periodic nature of the pseudorotation phase, polar plots are very
convenient for depicting five-membered ring conformations (see Figure 32). P will be
used as the angular coordinate and νmax will be used as the radial coordinate. Such
polar plots will be called pseudorotation plots or pseudorotation wheels [36].
3.2. The relationship between torsion angles and scalar couplings
As mentioned before, scalar couplings will be influenced by torsion angles, which is
the reason why conformational analysis can be performed by measuring coupling
constants. In 1959, Martin Karplus proposed a simple mathematical relationship
between 3JHH coupling constants and torsion angles (φ), based on the comparison of
experimental coupling constants and theoretical valence bond calculations [37].
Expressions relating scalar couplings and torsion angles are called Karplus relations,
named after its pioneer.
32
Figure 32 Pseudorotation plot with indicated positions of the 10 possible envelope and 10 possible twist conformations. This figure was taken from [38], with some small adaptations. With the used numbering scheme, the carbon carrying the B substituent is the β carbon, and the carbon with the OH substituent is the γ carbon.
𝐽 3 = 𝐴cos2(𝜑) + 𝐵 cos(𝜑) + 𝐶
The values for the A, B and C constants are derived empirically for specific
circumstances, for example, for certain substituents. Over the years, a lot of
refinements have been made with respect to this simple equation, mainly concerning
the effects of substituents, which have the largest impact on the scalar coupling next
to the torsion angle [36]. The substituent effects have been parameterized by the so-
called group electronegativities λi. These are used in the Diez-Donders equation, which
is a generalized Karplus equation for JHCCH couplings and which was used throughout
the course of this thesis [39, 40].
𝐽𝐻𝐶𝐶𝐻 = 𝐶0 + 𝐶1 cos(𝜑) + 𝐶2 cos(2𝜑) + 𝐶3 cos(3𝜑) + 𝑆2sin (2𝜑)
33
𝐶0 = 6.97 − 0.58∑𝜆𝑖𝑖
− 0.24(𝜆1𝜆2 + 𝜆3𝜆4)
𝐶1 = −1.06
𝐶2 = 6.55 − 0.82∑𝜆𝑖𝑖
+ 0.20(𝜆1𝜆4 + 𝜆2𝜆3)
𝐶3 = −0.54
𝑆2 = 0.68∑𝜉𝑖𝜆𝑖2
𝑖
Values for λ for various substituent groups have been deduced by Diez et al. ξ is 1 for
odd subscript values and -1 for even subscript values. As is evident from the equations,
C0 and C2, and thus JHCCH, are dependent on the substituent numbering scheme, which
can be seen in Figure 33.
Figure 33 Substituent numbering scheme used in the Diez-Donders equation. Figure taken from [36].
Marshall et al. studied the non-equivalence of exo-exo and endo-endo couplings in
norbornanes and found that non-bonded interactions can also have an influence on
scalar couplings [41]. This so-called Barfield effect can be explained from Figure 34.
The reduced distance between H1/H2 and X will reduce the 3JH1H2 scalar coupling,
however it will not affect the 3JH3H4 scalar coupling, nor any of the transoid couplings.
It has been proposed to correct for the Barfield effect by adding a correction term to
the Karplus relation.
∆𝐽 = 𝑇𝑏cos2(𝑃 − 𝑃𝑏)
Tb is dependent on the nature of atom X, and is 1 Hz for nitrogen and 2 Hz for carbon.
Pb is the pseudorotation phase for which the Barfield effect will be the most
pronounced. In case of the numbering scheme used, this will be 234° for α-β couplings,
90° or 270° for β-γ couplings (above or below the ring respectively), or 306° or 126°
for γ-δ couplings (above or below the ring respectively). Note that the Barfield
34
correction only needs to be applied for cisoid couplings and when |P – Pb| ≤ 90°, and
that it will always lead to a reduction of the scalar coupling [36].
Figure 34 Illustration of the Barfield effect in a five-membered ring. Figure taken from [36].
To relate 1H-19F couplings to H-C-C-F torsion angles, a Karplus relation that is also
parameterized with group electronegativities was used [42]. However, this relation also
needs bond angles (aFCC and aHCC) as parameters and obtaining sensible values for
those angles can be difficult as will be seen later.
𝐽𝐻𝐶𝐶𝐹 = 40.61cos2(𝜙) − 4.22 cos(𝜙) + 5.88 +∑𝜆𝑖[−1.27 − 6.20cos
2(𝜉𝑖𝜙 + 0.20𝜆𝑖)]
𝑖
− 3.72 (𝑎𝐹𝐶𝐶 + 𝑎𝐻𝐶𝐶
2− 110) cos2(𝜙)
Karplus relations deliver exocyclic torsion angles. However, the Altona-Sundaralingam
formalism, is built around endocyclic torsion angles. Luckily, the conversion between
exocyclic and endocyclic torsion angles is straightforward.
𝜈𝑒𝑥𝑜 = 𝐴𝜈𝑒𝑛𝑑𝑜 + 𝐵
For an ideal tetrahedral geometry, A will be 1 and B will be 0° for cisoid couplings or
±120° for transoid couplings. When deviating from the ideal tetrahedral geometry,
small corrections have to be applied to A and B. However, in this thesis, ideal
tetrahedral geometry was always assumed [36].
3.3. Fitting conformations
Two fitting programs were used, a graphical user interface (GUI), written by P.M.S.
Hendrickx [43], and an in-house MatLab script, written by D. Sinnaeve. Initially, the
GUI was used. However, due to some shortcomings of the GUI, the new MatLab script
was written. The fitting procedure in this script follows the same principle as the one
used in the GUI, with the advantages that the known coding of the algorithm provides
full control, 1H-19F couplings can be used (using the Karplus relation of Thibaudeau et
35
al. [42]), and the Barfield correction can optionally be applied. A comparison between
the GUI and the MatLab script is made in the supporting information (section 6.7.1).
3.3.1. MatLab script
To derive one conformation from a set of experimental scalar couplings, the fitting
program works in the following way. Starting from initial guesses for P and νmax,
endocyclic torsion angles are calculated via the Altona-Sundaralingam formalism. The
obtained endocyclic torsion angles are then converted to exocyclic torsion angles
between the protons for which the coupling constants were measured. These exocyclic
torsion angles are then fed into the Diez-Donders equations, together with the correct
λ parameters, to yield the estimated scalar coupling constants. These are
subsequently compared with the experimental coupling constants by calculating the
root-mean-squared difference (RMSD) as goodness-of-fit.
𝑅𝑀𝑆𝐷 = √1
𝑁∑(𝐽𝑖
𝑐𝑎𝑙𝑐 − 𝐽𝑖𝑒𝑥𝑝𝑡)2
𝑁
𝑖
An optimization procedure is used to update the initial P and νmax parameters in order
to minimize the RMSD. These updated parameters are compared to the original
parameters and convergence is checked. Iteration of the previously described steps is
performed until convergence is reached [36].
An equilibrium between two conformations can also be fitted, but then there are some
differences. Five parameters now need to be fitted, so initial values have to be entered
for PA, PB, νmax,A, νmax,B and xA. The latter is the fraction of conformer A, from which the
fraction of conformer B follows, since xA + xB = 1. All the previously described steps up
until the calculation of the coupling constants are now performed for two conformers
separately. Then, an extra step is added in which weighted average coupling constants
are calculated, using the following formula.
𝐽𝑖𝑐𝑎𝑙𝑐 = 𝑥𝐴𝐽𝑖
𝐴 + 𝑥𝐵𝐽𝑖𝐵
Now these calculated scalar coupling constants are used to calculate the RMSD. The
optimization procedure remains the same [36, 43].
36
The MatLab script also allows estimating the impact of the precision of the coupling
measurement on the fitting. This is done using a Monte Carlo analysis [44]. In this
procedure, within 200 (one conformer) or 500 (two conformers) iterations, Gaussian
noise with a spread equal to a precision of 0.1 Hz (see supporting information, section
6.7.2) is added to the experimental coupling values in a semi-random fashion. These
modified coupling constants are then used to fit the conformation(s), so in fact 200 or
500 conformations will be fitted. This multitude of fittings, will be used for two purposes.
First, they allow the calculation of 95% uncertainties of the fitted parameters. Second,
the results from the 200/500 different one/two conformations can be plotted in a
pseudorotation plot, providing a visual indication of the uncertainty on the fitted
parameters due to the measurement precision.
3.3.2. Approximations and limitations
It is important to realize that, even when the group electronegativities in the generalized
Diez-Donders equations are known, a Karplus equation will always be an approximate
description of the relation between torsion angle and scalar coupling, and it is thus
limited in its accuracy. For the fluoroproline systems, further complications arise that
will negatively affect the accuracy of the Karplus relations. First, not all group
electronegativity values required, have been reported in literature. This means that for
some substituents, approximations for this value had to be made (see supporting
information for details), giving the Karplus relation an even more approximate nature.
Furthermore, bond angles (aFCC and aHCC) are needed as input for the 1H-19F Karplus
relation. Fluorination can indeed significantly perturb the ideal tetrahedral bond angles,
and this is not easy to estimate. A solution would be resorting to crystal structures,
however, only for Ac-(4R)-FPro-OMe has this been reported in literature [45]. For the
other fluorinated analogues results from quantum chemical calculations had to be
used. These were kindly provided by dr. Ilya Kuprov (Southampton University). Using
the Gaussian program, he performed geometry optimizations for the four molecules
with DFT, using the M06 functional on the cc-pVDZ level of theory. From this, sensible
values for the required bond angles could be obtained, which are listed in the
supporting information.
37
3.4. Results and discussion
Because the results obtained via both the GUI and the MatLab script (when 1H-19F
couplings are not used) are very similar, only the results from the MatLab script will be
discussed here. All fit results (pseudorotation plots, fitted parameters, 95%
uncertainties, RMSD values and residuals) can be found in the supporting information
(sections 6.7.4 and 6.7.5). The input parameters used can also all be found in the
supporting information. For obtaining initial P and νmax values, a preliminary scan of
the parameter space was performed, so the actual fitting can start close the RMSD
minimum.
3.4.1. Ac-Pro-OMe
For Ac-Pro-OMe, no fitting could be performed for the trans conformer in D2O as only
two couplings could be measured, due to strong coupling artefacts (see Chapter 2).
The cis conformer in CDCl3 is also plagued by strong coupling artefacts, but here
averages of couplings could be measured, so fitting could be performed. However,
results for this particular case have to be interpreted with caution.
When trying to fit one conformation, a single well-defined conformation appears to be
obtained only for the cis conformer in CDCl3. The P value is 175.7°, νmax is 17.3° and
the RMSD is only 0.6 Hz, which seems to indicate a rather good result. The fitted
conformation is close to a Cγ endo conformation, which is the predominant conformer
for Ac-Pro-OMe according to literature [1]. However, it is reported that Ac-Pro-OMe
does not exclusively adopt a Cγ endo conformation, but rather a 7:3 ratio of Cγ endo
and Cγ exo [1]. In that respect, it is unexpected that fitting one conformation to the data
yields such a low RMSD. In contrast, in case of the trans conformer in CDCl3 and the
cis conformer in D2O, fitting one conformation to the data does not yield a reliable
result, as the uncertainties on the P value are of the same order of magnitude as the
P value and the uncertainties on νmax are even five to six orders of magnitude larger
than vmax. The RMSD values are also quite high for both cases (2.6 Hz and 3.2 Hz
respectively). These observations all indicate that assuming one conformation for Ac-
Pro-OMe in these cases cannot adequately explain the observed scalar couplings, and
that indeed a second conformer should be assumed. The fact that a single
conformation could adequately fit the data of the cis conformer in CDCl3, could be
caused by the high proportion of average couplings used. This significantly reduces
38
the restrictiveness on the system imposed by the experimental data, which might make
it easier to fit one conformation. It would indeed be very surprising if such a significant
difference in ring pucker equilibrium between the cis and trans form in chloroform would
be reflected by only such limited differences in coupling constants between the two
forms (see Table 7 supporting info).
When assuming two conformers, the results shown in Figure 36 are obtained. Again,
the fitting for the cis conformer in CDCl3 proves to be less reliable, as it leads to rather
ill-defined conformations. For the other two cases, the RMSD values have decreased
compared to fitting one conformation (to 0.5 Hz and 0.8 Hz). The fitted values for trans
Ac-Pro-OMe in CDCl3 were P = 187.9°, νmax = 31.5° for the major pucker, and P =
17.1°, νmax = 49.7° for the minor pucker, with a fraction of the minor pucker of 29.5%.
Figure 35 Pseudorotation plots of Ac-Pro-OMe for fitting one conformer. Left: CDCl3. Right: D2O. Above: trans conformer. Below: cis conformer.
39
The fitted values for cis Ac-Pro-OMe in D2O were P = 182.8°, νmax = 36.4° for the major
pucker, and P = 13.4°, νmax = 43.7° for the minor pucker, with a fraction of the minor
pucker of 16.9%. From Figure 36, one can clearly see that the uncertainty on the νmax
value of the minor pucker is quite large in all cases, which is not very surprising. Since
this conformer is only present as a small fraction, the weighted average couplings
calculated in the script will only vary slightly, even for larger variations of
pseudorotation parameters of the minor fraction. These obtained values very much
confirm what has been already found in literature (vide supra), as a Cγ endo and Cγ
exo conformer correspond to P values of 198° and 18° respectively. It can also be
noted that the Cγ exo fraction is larger for the trans conformer than for the cis
conformer, which can be expected based on the coupled equilibria described in
Chapter 1.
Figure 36 Pseudorotation plots of Ac-Pro-OMe for fitting two conformers. Left: CDCl3. Right: D2O. Above: trans conformer. Below: cis conformer.
40
3.4.2. Ac-(4R)-FPro-OMe
Fitting was performed with a maximum of six 1H-1H couplings and four 1H-19F
couplings. However, this maximum number of couplings could not be reached for the
trans conformer in CDCl3, as two 1H-1H couplings were measured as an average and
two 1H-19F couplings could not be measured.
When fitting one conformer without using the 1H-19F couplings, the pseudorotation
plots in Figure 37 are generated. From these plots, it can be seen that rather well-
defined puckers are obtained. In all cases, similar ring puckers are found with
pseudorotation phases ranging from 13.5° to 25.0° and puckering amplitudes ranging
from 40.7° to 48.0°. RMSD values all fall around 0.4 Hz, except for the trans conformer
in CDCl3, where 0.94 Hz is obtained. This might be related to the fact that not all
individual couplings were known in this case (vide supra). Nevertheless, these are
quite low RMSD values, which indicates a high goodness-of-fit. The low uncertainty
values on the fitted parameters also indicate that the puckering equilibrium is skewed
to one side, because if a second conformer was present with reasonable percentages,
the uncertainties are typically larger, as was seen for Ac-Pro-OMe. According to
literature [1], the ring pucker of Ac-(4R)-FPro-OMe is indeed expected to show a large
preference for the Cγ exo conformer, which has a pseudorotation phase of 18°. The
results from the fitting thus confirm the presence of this conformer, and its
predominance. It can be noted that the νmax values are slightly higher than for Ac-Pro-
OMe.
41
Figure 37 Pseudorotation plots of Ac-(4R)-FPro-OMe for fitting one conformer, without use of 1H-19F couplings. Left: CDCl3. Right: D2O. Above: trans conformer. Below: cis conformer.
If one includes 1H-19F couplings in the conformational analysis, one sees that roughly
the same results are obtained, which would be expected. However, the RMSD values
are systematically higher (for all but one, more than 1 Hz), which may point out that
the 1H-19F Karplus relation is not as accurate as those for 1H-1H, since the highest
residuals were often found for 1H-19F couplings (all values are in the supporting
information, as well as the pseudorotation plots, sections 6.7.4 and 6.7.5).
In order to fit two conformations, 1H-19F couplings are needed as input as six 1H-1H
couplings are insufficient to fit five parameters (see supporting information, section
6.7.1). For the trans conformers in both solvents, the uncertainty on the fitted fraction
of the minor pucker is so large (much larger than the fitted value of the fraction), that it
is impossible to say that there are two conformations. For the cis conformers in both
solvents, the uncertainty of the fraction is smaller, however the fitted νmax values for
the minor conformer are so unreasonably high (104.0° in CDCl3 and 147.2° in D2O)
that these would not be realistic conformations (some endocyclic torsion angles would
be over 100°). For these cases, the major conformer coincides rather well with the Cγ
exo conformation, and the uncertainties on its P and νmax values are very small. This
also indicates that it is not possible to prove the presence of a second conformer,
because the large uncertainties of the minor pucker imply that this pucker does not
contribute much to the eventual coupling constants. It is reasonable to think that there
probably will exist a puckering equilibrium with a small fraction of Cγ endo. However,
with the limited number of couplings and the approximations used in the Karplus
42
relations, it does not appear possible to detect this. If we want to be able to confirm
this hypothesis, more accurate Karplus relations or additional experimental data will
be needed. Pseudorotation plots and all fitted parameters are in the supporting
information (section 6.7.4 and 6.7.5).
3.4.3. Ac-(4S)-FPro-OMe
The trans conformer in D2O is expected to yield less reliable results, as two pairs of
1H-1H couplings had to be measured as averages, and two 1H-19F couplings could not
be measured. For the cis conformer in D2O, also two 1H-19F couplings could not be
measured.
When looking at Figure 38, which are the plots for fitting one conformation without use
of the 1H-19F couplings, one can see that three of the four fittings yield approximately
the same conformation. Only the trans conformer in D2O, yields another conformation,
but this is less well-restrained due to use of average coupling values. Discarding, this
result, one can see that the other results have values for P ranging from 199.3° to
206.5° and values for νmax ranging from 37.1° to 41.6°. The uncertainties due to
measurement errors are a bit higher than for Ac-(4R)-FPro-OMe. Nonetheless, also for
this molecule it can be deduced that there will be a large preference for one conformer.
According to literature [1], this should be the Cγ endo conformation, which has a P
value of 198°. This is indeed in good agreement with the fitted results. When including
1H-19F couplings (again discarding the results for the trans conformer in D2O), the P
values shift to lower values for both cases in CDCl3 (179.4° for trans and 175.8° for
cis), which is more a twist conformation (γTβ, which means Cγ is below the plane and
Cβ above), but nevertheless still close to 198°. Just like for Ac-(4R)-FPro-OMe, the
RMSD values are higher upon inclusion of 1H-19F couplings (see supporting info).
When fitting two conformations (necessitating the use of 1H-19F couplings, vide supra),
the story for Ac-(4S)-FPro-OMe is very similar as for Ac-(4R)-FPro-OMe, although now
for all four cases, the uncertainties on the fitted fractions were much higher than the
values of the fitted fractions themselves. This again indicates that the presence of a
second conformer cannot be proven with the currently used approximations. A similar
conclusion as for Ac-(4R)-FPro-OMe can thus be drawn, but now with the Cγ exo
conformer as the minor fraction. Again, the pseudorotation plots can be found in the
supporting information (section 6.7.5).
43
Figure 38 Pseudorotation plots of Ac-(4S)-FPro-OMe for fitting one conformer without use of 1H-19F couplings. Left: CDCl3. Right: D2O. Above: trans conformer. Below: cis conformer.
3.4.4. Ac-4,4-F2Pro-OMe
No fitting could be performed for Ac-4,4-F2Pro-OMe without using 1H-19F couplings as
the number of vicinal 1H-1H couplings (two) is insufficient. For the cis conformer in
CDCl3, one 1H-19F coupling could not be measured and for the trans conformer in D2O,
two pairs of 1H-19F couplings were measured as averages.
When fitting one conformation, except for the trans conformer in CDCl3, well-defined
conformations could be fitted, which for the two cis conformers are close to a Cγ endo
conformation (P = 167.7°; νmax = 21.3°, in CDCl3 and P = 188.1°; νmax = 29.4°, in D2O).
For the trans conformer in D2O, a rather unexpected conformation is fitted (P value is
133.9°). However, the reliability of this result is reduced because of the featured
44
average couplings in the dataset. According to literature [4], Ac-4,4-F2Pro-OMe should
be close to Ac-Pro-OMe in terms of conformational behaviour, with a predominant Cγ
endo conformation, but also a considerable fraction of Cγ exo. Just like for Ac-Pro-
OMe, high RMSD values are obtained (even for both cis conformers, 4.1 Hz in CDCl3
and 2.7 Hz in D2O), which indeed suggests that the assumption of one conformer is
inadequate.
Figure 39 Pseudorotation plots of Ac-4,4-F2Pro-OMe for fitting one conformation, using 1H-19F couplings. Left: CDCl3. Right: D2O. Above: trans conformer. Below: cis conformer.
45
Figure 40 Pseudorotation plots of Ac-4,4-F2Pro-OMe for fitting two conformations, using 1H-19F couplings.. Left: CDCl3. Right: D2O. Above: trans conformer. Below: cis conformer.
When fitting two conformations (Figure 40), two rather well-defined conformations can
always be fitted for Ac-4,4-F2Pro-OMe. However, the fitted conformations are
inconsistent for the four different cases, and are also not close to either a Cγ endo or
Cγ exo conformation. Furthermore, the νmax values of the minor conformers are always
unreasonably high.
The observations in literature concerning the ring pucker of Ac-4,4-F2Pro-OMe (vide
supra) are based on the assumption that the two opposing stereoelectronic effects of
the fluorine atoms will completely cancel each other out. Renner et al. [4], drew this
conclusion by comparing the patterns of 1H-1H and 1H-19F coupling constants of
different proline analogues, without attempting an extensive conformational analysis
as presented here. They reasoned that comparable patterns should yield comparable
46
conformations and from this they concluded that Ac-4,4-F2Pro-OMe should be
predominantly Cγ endo. It is clear that the possibility of other ring puckers than Cγ endo
or Cγ exo was never considered. It is thus not impossible that different conformations
are significantly populated, and that this is reflected in our conformational analysis.
However, 8 out of 10 used couplings were 1H-19F couplings, for which the 1H-19F
Karplus relation needs to be used, which is a relation of which the accuracy for the
fluoroproline analogues is not fully established, considering also the approximations
used for group electronegativities and bond angles. It could be that other
parameterizations than the ones used now are needed. Therefore, it can be concluded
that further research into refining the Karplus relations will be needed to draw a
confident conclusion concerning Ac-4,4-F2Pro-OMe.
If it is assumed, that only the Cγ endo or Cγ exo conformations are possible (with the
values obtained from Ac-(4S)-FPro-OMe, P = 205.2°, νmax = 36.8°; and Ac-(4R)-FPro-
OMe, P = 26.8°, νmax = 50.0°, respectively), the experimental couplings of the Ac-4,4-
F2Pro-OMe samples are used to fit the fraction based on the fact that the lowest RMSD
value gives the best fit. Except for the trans conformer in D2O, where insufficient
individual couplings could be measured, such an analysis was performed. Consistently
it was found that the Cγ exo conformer is more predominant than the Cγ endo
conformer. This contradicts the literature assumption that Ac-4,4-F2Pro-OMe behaves
similarly as Ac-Pro-OMe, so still it can be concluded that further research is required.
However, care has to be taken not to attach too much importance to these results, as
even the minimal RMSD value is quite high, 2.8 Hz (plots can be found in the
supporting information, section 6.7.6).
3.4.5. Fitting with the Barfield correction
In all previous results, no Barfield corrections were taken into account. The analyses
were repeated on those samples for which (nearly) all individual 1H-1H couplings could
be measured. For all cases except two, the Barfield correction gave similar to equal
results, but with higher or equal RMSD values. For the cis conformer of Ac-4,4-F2Pro-
OMe and Ac-(4R)-FPro-OMe both in CDCl3, completely different puckers were
obtained. However, in those cases the RMSD values were a lot higher than before
applying the Barfield correction. It can be concluded that the proposed Barfield
correction does not provide an added value for the investigated samples. All results
can be found in the supporting information (sections 6.7.4 and 6.7.5).
47
4. Thermodynamics and kinetics of the cis/trans isomerism
The phenomenon of cis/trans isomerism was already described in Chapter 1, and in
Chapter 2 it was explained that this kind of isomerism also takes place in the
investigated molecules. In this chapter, both the thermodynamics and kinetics of the
cis/trans isomerism will be discussed.
4.1. Investigation of the thermodynamics
4.1.1. Theory
The central property in chemical thermodynamics is the equilibrium constant K. In case
of the investigated equilibrium (Figure 41) it can be described as follows.
𝐾𝑡𝑟𝑎𝑛𝑠/𝑐𝑖𝑠 =[𝑡𝑟𝑎𝑛𝑠]
[𝑐𝑖𝑠]
In which [trans] and [cis] are the equilibrium concentrations of the trans and cis
conformer respectively. The equilibrium constant can be related to the standard Gibbs
free energy of the reaction ΔG°trans/cis by the following formula.
𝐾𝑡𝑟𝑎𝑛𝑠/𝑐𝑖𝑠 = 𝑒−∆𝐺𝑡𝑟𝑎𝑛𝑠/𝑐𝑖𝑠
0
𝑅𝑇
In order to retrieve the reaction enthalpy ΔH°trans/cis and the reaction entropy ΔS°trans/cis,
this equation has to be rearranged.
∆𝐻𝑡𝑟𝑎𝑛𝑠/𝑐𝑖𝑠0 − 𝑇∆𝑆𝑡𝑟𝑎𝑛𝑠/𝑐𝑖𝑠
0 = −𝑅𝑇 ln (𝐾𝑡𝑟𝑎𝑛𝑠/𝑐𝑖𝑠)
If -RTln(K) is plotted versus temperature, the intercept will be equal to the reaction
enthalpy and the negative of the slope will be equal to the reaction entropy (see
supporting information section 6.8.2).
Figure 41 The cis/trans isomerism for Ac-Pro-OMe.
48
Ktrans/cis can be easily measured from quantitative 1D spectra, since the intensities of
the peaks should reflect the concentrations of both forms. So for a certain resonance
i, Ktrans/cis can also be written as follows.
𝐾𝑡𝑟𝑎𝑛𝑠/𝑐𝑖𝑠 =𝐼𝑖𝑡𝑟𝑎𝑛𝑠
𝐼𝑖𝑐𝑖𝑠
Resonances fit to provide intensities by integration, should be resolved from other
resonances. Typically, either one of the methyl resonance pairs in the 1D 1H spectrum,
or the 19F resonances in case of the monofluorinated compounds were used, except
for Ac-(4R)-FPro-OMe, where a resolved β 1H multiplet had to be used. Spectra were
measured at six different temperatures in the case of CDCl3, and three in D2O. The
spectra of the CDCl3 samples were measured non-quantitatively, however, this is
justifiable, as T1 relaxation for a certain resonance will be approximately the same for
the cis as for the trans conformer (see supporting information, section 6.8.1).
4.1.2. Results and discussion
The results for Ktrans/cis are presented in Tables 2 and 3. Three trends can be seen.
Firstly, Ktrans/cis drops with increasing temperature. According to Le Châtelier’s
principle, when temperature is increased and the equilibrium shifts in the opposite
direction, it implies that the reaction is exothermic. This confirms that the trans
conformer is more energetically stable than the cis conformer, as already discussed in
Chapter 1. Secondly, Ac-(4R)-FPro-OMe has the highest values for Ktrans/cis while Ac-
(4S)-FPro-OMe has the lowest values, while the values for Ac-Pro-OMe and Ac-4,4-
F2Pro-OMe fall in-between. It thus appears that the cis/trans equilibrium follows a
similar trend as the ring pucker. As discussed in Chapter 1, it is known that a stabilizing
n → π* interaction is present in the trans form when a Cγ exo pucker is adopted,
providing an explanation for this correlation. As shown in Chapter 3, Ac-(4R)-FPro-
OMe has a large preference for the Cγ exo pucker, which shifts the equilibrium more
towards the trans conformer. Ac-(4S)-FPro-OMe, on the other hand, has a large
preference for the Cγ endo pucker, which will shift the cis/trans equilibrium in the
opposite direction. However, Ktrans/cis is still larger than 1, so the trans conformer is still
more prevalent. The puckering equilibrium of Ac-4,4-F2Pro-OMe is not really biased
towards one form, which explains why its Ktrans/cis values are closer to those of Ac-Pro-
OMe. Lastly, the Ktrans/cis values in CDCl3 are always lower than those in D2O. A
49
possible explanation for this was provided by Siebler et al. [46]. They theorized that
the increased charge separation induced by the n → π* interaction between the
adjacent carbonyl groups is better solvated and therefore more stabilized in D2O. Since
these interactions only occur in the trans form, this can be a reason why the population
of the trans form is higher in D2O than in CDCl3.
Table 2 Ktrans/cis values for the CDCl3 samples at different temperatures.
T(K) Ac-Pro-OMe Ac-(4R)-FPro-OMe Ac-(4S)-FPro-OMe Ac-4,4-F2Pro-OMe
283 4.36 4.12 1.80 3.29
288 4.22 4.48 1.80 3.18
293 4.03 4.32 1.74 3.10
298 3.89 4.20 1.70 3.03
303 3.76 4.00 1.68 2.95
308 3.67 3.89 1.65 2.92
Table 3 Ktrans/cis values for the D2O samples at different temperatures.
T(K) Ac-Pro-OMe Ac-(4R)-FPro-OMe Ac-(4S)-FPro-OMe Ac-4,4-F2Pro-OMe
298 5.04 7.23 2.50 3.51
308 4.79 6.44 2.41 3.33
333 4.03 5.19 2.18 3.02
As described above, ΔH°trans/cis and ΔS°trans/cis values can be obtained from these data
by linear regression (Table 4). The negative values for ΔH°trans/cis point out that the
reaction is exothermic, and the absolute values more or less show the same trends
between different molecules and different solvents as could be seen for Ktrans/cis. The
values for Ac-4,4-F2Pro-OMe in D2O are not very reliable, since the regression is not
good (R² = 0.89). The values for ΔS°trans/cis are all negative, which indicates that the
trans conformer is more ordered than the cis conformer. A possible rationalization can
be found when considering the n → π* interactions (see Chapter 1). These interactions
can only occur in the trans conformer, and will limit the atomic motion, making the
molecules more rigid and ordered.
Table 4 ΔH°trans/cis (in kJ/mol) and ΔS°trans/cis (in J/mol K) values for the investigated molecules in both solvents.
Ac-Pro-OMe Ac-(4R)-FPro-OMe Ac-(4S)-FPro-OMe Ac-4,4-F2Pro-OMe
CDCl3 D2O CDCl3 D2O CDCl3 D2O CDCl3 D2O
ΔH°trans/cis -5.1 ± 0.4 -5.4 ± 2.9 -5.3 ± 0.6 -7.7 ± 2.0 -2.5 ± 0.2 -3.2 ± 0.3 -3.8 ± 0.3 -3.5 ± 0.9
ΔS°trans/cis -5.9 ± 1.3 -4.5 ± 9.2 -5.9 ± 2.1 -9.6 ± 6.5 -4.0 ± 0.8 -3.2 ± 1.0 -3.5 ± 1.0 -1.3 ± 2.9
50
4.2. Investigation of the kinetics
4.2.1. Theory
Consider a system whereby a spin is in equilibrium between two identities I and S with
an equilibrium constant K, a forward reaction rate constant k1 and a backward reaction
rate constant k-1.
𝐼 ⇋ 𝑆
When an equal relaxation rate constant R1 is assumed in both identities – which is
justifiable, see the results of the inversion recovery experiments in the supporting
information, section 6.8.1 – the deviation of the magnetizations of the spins will evolve
according to the following equations.
{
𝑑∆𝐼(𝑡)
𝑑𝑡= −(𝑅1 + 𝑘1)∆𝐼(𝑡) + 𝑘−1∆𝑆(𝑡)
𝑑∆𝑆(𝑡)
𝑑𝑡= −(𝑅1 + 𝑘−1)∆𝑆(𝑡) + 𝑘1∆𝐼(𝑡)
This can also be written in matrix representation:
𝑑
𝑑𝑡(∆𝐼(𝑡)
∆𝑆(𝑡)) = (
−(𝑅1 + 𝑘1) 𝑘−1𝑘1 −(𝑅1 + 𝑘−1)
) (∆𝐼(𝑡)
∆𝑆(𝑡))
Which abbreviated is:
𝑑
𝑑𝑡∆𝐌(𝑡) = [𝐑 + 𝐊]∆𝐌(𝑡)
(4.1)
Since [R + K] is independent of time, the general solution to this differential equation
is as follows.
∆𝐌(𝑡) = 𝑒[𝐑+𝐊]𝑡∆𝐌(0)
Applying the matrix exponential to [R + K]t yields:
𝑒[𝐑+𝐊]𝑡 =
(
𝑘1𝑘1 + 𝑘−1
𝑒−(𝑅1+𝑘1+𝑘−1)𝑡 +𝑘−1
𝑘1 + 𝑘−1𝑒−𝑅1𝑡 −
𝑘−1𝑘1 + 𝑘−1
𝑒−(𝑅1+𝑘1+𝑘−1)𝑡 +𝑘−1
𝑘1 + 𝑘−1𝑒−𝑅1𝑡
−𝑘1
𝑘1 + 𝑘−1𝑒−(𝑅1+𝑘1+𝑘−1)𝑡 +
𝑘1𝑘1 + 𝑘−1
𝑒−𝑅1𝑡𝑘−1
𝑘1 + 𝑘−1𝑒−(𝑅1+𝑘1+𝑘−1)𝑡 +
𝑘1𝑘1 + 𝑘−1
𝑒−𝑅1𝑡
)
The following relation exists between the equilibrium constant, the equilibrium
populations p of I and S, and the kinetic constants:
51
𝐾 =𝑝𝑆𝑝𝐼=𝑘1𝑘−1
Thus, it can be written that:
𝑝𝑆 = 𝑝𝐼𝑘1
𝑘−1 or 𝑝𝐼𝑘1 = 𝑝𝑆𝑘−1
And from the definition of kex = k1 + k-1, and with pI + pS normalized to 1, it follows that:
{𝑝𝐼𝑘𝑒𝑥 = 𝑝𝐼𝑘1 + 𝑝𝐼𝑘−1 = 𝑝𝑆𝑘−1 + 𝑝𝐼𝑘−1 = 𝑘−1𝑝𝑆𝑘𝑒𝑥 = 𝑝𝑆𝑘1 + 𝑝𝑆𝑘−1 = 𝑝𝑆𝑘1 + 𝑝𝐼𝑘1 = 𝑘1
With these relations the exponential of [R + K]t can be rewritten:
𝑒[𝐑+𝐊]𝑡 = (
(𝑝𝐼 + 𝑝𝑆𝑒−𝑘𝑒𝑥𝑡)𝑒−𝑅1𝑡 𝑝𝐼(1 − 𝑒
−𝑘𝑒𝑥𝑡)𝑒−𝑅1𝑡
𝑝𝑆(1 − 𝑒−𝑘𝑒𝑥𝑡)𝑒−𝑅1𝑡 (𝑝𝑆 + 𝑝𝐼𝑒
−𝑘𝑒𝑥𝑡)𝑒−𝑅1𝑡)
(4.2)
These equations, (4.1) and (4.2), form the basis to derive the outcome of the selective
inversion recovery and the EXSY experiments.
4.2.1.1. Selective inversion recovery equation
This experiment was introduced in Chapter 2. When doing a selective inversion
recovery, where only the resonance from S is inverted while that of I remains
unaffected, the deviations in z-magnetizations immediately after the 180° pulse (at time
0) will be the following.
{
∆𝐼(0) = 0
∆𝑆(0) = −2𝑆0 = −2𝑝𝑆𝑀0 = −2𝑝𝐼𝑀
0𝑘1𝑘−1
With M0 the total equilibrium z-magnetization of spins I and S (with I0 = pIM0 and S0 =
pSM0). When plugging these values in equation (4.1) (with (4.2) as the exponential of
[R + K]t), a description of the evolution of the perturbation of the z-magnetization of
spin I as a function of time is obtained.
𝛥𝐼(𝑡) = −2𝑝𝐼𝑀0(𝑝𝐼𝑒
−𝑅1𝑡 − 𝑝𝐼𝑒−(𝑅1+𝑘𝑒𝑥)𝑡)
𝑘1𝑘−1
This result reveals that, after selective inversion of S, the z-magnetisation of I will first
decease with an initial rate of kex, then reach a minimum, and finally return to its
52
equilibrium value due to T1- relaxation (see Figure 42). Applying a 90° pulse at time t
will convert the z-magnetization into detectable signal, the integral of which will abide
the above expression. Measurement of spectra at different values of t, thus allows
extraction of kex through curve fitting. Since, in practice, the selective inversion is
imperfect, and the magnetization evolves before detection, a slightly modified equation
is used to perform the data fitting (in which I(t) is used instead of ΔI(t)).
𝐼(𝑡) = 𝐼0 (1 − (𝑎𝑒−𝑡𝑇1 − 𝑏𝑒
−𝑡(1𝑇1+𝑘𝑒𝑥)))
In which kex, I0, a and b are the parameters that are being fitted, and in which T1 (=
1/R1) is required as input. T1 constants for all samples were measured with inversion
recovery experiments (see Chapter 2) and fitted values can be found in the supporting
information (section 6.8.1).
Figure 42 Selective inversion recovery of Ac-(4R)-FPro-OMe in CDCl3 at 308 K.
4.2.1.2. EXSY equations
The ΔM(0) matrix, at the start of the mixing time and after the t1-evolution, is given by:
∆𝐌(0) = (
−𝑝𝐼𝑀0 cos𝜔𝐼𝑡1
−𝑝𝑆𝑀0 cos𝜔𝑆𝑡1
) (4.3)
With ω the Larmor frequency. This means that for spin I, the peak intensity will evolve
as (plugging in (4.2) and (4.3) in (4.1)):
−𝑀0 cos𝜔𝐼𝑡1 [𝑝𝐼(𝑝𝐼 + 𝑝𝑆𝑒−𝑘𝑒𝑥𝑡)𝑒−𝑅1𝑡] − 𝑀0 cos𝜔𝑆𝑡1 [𝑝𝐼𝑝𝑆(1 − 𝑒
−𝑘𝑒𝑥𝑡)𝑒−𝑅1𝑡]
The cos ωIt1 part represents the diagonal peak and the cos ωSt1 part represents the
cross-peak. A similar expression can be found for spin S:
53
−𝑀0 cos𝜔𝑆𝑡1 [𝑝𝑆(𝑝𝑆 + 𝑝𝐼𝑒−𝑘𝑒𝑥𝑡)𝑒−𝑅1𝑡] − 𝑀0 cos𝜔𝐼𝑡1 [𝑝𝐼𝑝𝑆(1 − 𝑒
−𝑘𝑒𝑥𝑡)𝑒−𝑅1𝑡]
In this case, the cos ωSt1 part represents the diagonal peak and the cos ωIt1 part
represents the cross-peak.
To find kex, the ratio of the integrals of the diagonal peak and the cross-peak at the
same F2 frequency, has to be taken:
𝑟𝐼 =𝐼𝐼𝑐𝑟𝑜𝑠𝑠
𝐼𝐼𝑑𝑖𝑎𝑔
=𝑝𝑆(1 − 𝑒
−𝑘𝑒𝑥𝑡)
𝑝𝐼 + 𝑝𝑆𝑒−𝑘𝑒𝑥𝑡
An analogous expression can be found for spin S. With some rearrangement, the
following equations for kex can be obtained:
𝑘𝑒𝑥 = −
ln(1 −
𝑝𝐼𝑝𝑆𝑟𝐼
1 + 𝑟𝐼)
𝑡𝑚
𝑘𝑒𝑥 = −
ln(1 −
𝑝𝑆𝑝𝐼𝑟𝑆
1 + 𝑟𝑆)
𝑡𝑚
With tm the EXSY mixing time. Via these formulae, values for kex could be determined
from 2D EXSYs. These were measured at various mixing times to check the
consistency of the results, and which also provided an estimation of error for kex. The
signals used to determine kex, both for the selective inversion recovery and the EXSY,
were the 19F signals in case of the monofluorinated samples, or the (resolved) methyl
1H signals for the other samples.
4.2.1.3. Arrhenius and Eyring equations
When the values for kex are determined for different temperatures, the activation
energy of the reaction (Ea) and the pre-exponential factor (A) can be determined by
using the Arrhenius equation.
𝑘 = 𝐴𝑒−𝐸𝑎𝑅𝑇
The k in this equation is not kex, but rather k1 or k-1, so for both the forward and
backward reaction, Ea and A can be determined (k1 and k-1 can be derived from kex if
K is known, vide supra). In order to obtain Ea and A, the Arrhenius equation is
linearized:
54
ln(𝑘) = ln(𝐴) −𝐸𝑎𝑅𝑇
The values of ln(k) are plotted as a function of 1/RT and linear regression is performed.
The slope will be equal to -Ea and the intercept will be equal to ln(A). The resulting
graphs can be found in the supporting information (section 6.8.3).
The Eyring equation is similar to the Arrhenius equation and it relates k1 or k-1 to the
enthalpy of activation ΔH‡ and the entropy of activation ΔS‡.
𝑘 = 𝑘𝐵𝑇
ℎ𝑒∆𝑆‡
𝑅 𝑒−∆𝐻‡
𝑅𝑇
The linear form of this equation is as follows:
𝑅 ln (ℎ𝑘
𝑘𝐵𝑇) = −∆𝐻‡
1
𝑇+ ∆𝑆‡
With kB the Boltzmann constant and h Planck’s constant. The values of 𝑅 ln (ℎ𝑘
𝑘𝐵𝑇) are
plotted versus 1/T. Linear regression will yield -ΔH‡ as the slope and ΔS‡ as the
intercept. The resulting graphs can be found in the supporting information (section
6.8.4).
4.2.2. Results and discussion
In first instance, only the selective inversion recovery was used to determine the kex
values, as it was expected that this strategy would allow a faster sampling of various
mixing times. However, this approach had some issues. First, a selective pulse (Iburp)
was used for excitation. Such a pulse is quite long, meaning that relaxation and
exchange will already take place during this time, before the start of acquisition. When
kex is small compared to the inverse of the pulse duration, this may result in a situation
where the sampling of the curve starts at a position that is too late for reliably fitting kex
values. Especially at low temperatures this was the case, or when the signals of I and
S were too close to each other, necessitating a long duration of the selective pulse. An
alternative approach was then used for selective inversion, using a 90°x — Δ — 90°–x
sequence, with Δ a delay equal to half of the inverse of the frequency difference
between I and S, and the rf-frequency set to the resonance frequency of spin I (see
Chapter 2). This approach, which only works for singlets, is generally shorter than a
selective pulse, making it easier to fit the function to the data. However, there were still
55
cases where kex turned out to be too small (i.e. in D2O at 308 K, or most Ac-Pro-OMe
samples), so the fitting is still problematic.
Finally, we resorted to 2D EXSY to determine kex. For Ac-Pro-OMe in CDCl3 at 308 K,
where the selective inversion recovery experiments worked out, it was validated that
the same result was obtained with EXSY. For this reason, all other cases where the
selective inversion recovery experiments were already measured and yielded reliable
fitting, no EXSY experiments (which takes ca. 4.5 h per sample) were measured
additionally. When the selective inversion recovery was used, 95% errors (ε95) were
calculated using a Monte Carlo analysis [44], similar to the one described in Chapter
3. When the EXSY was used, 95% errors were calculated based on the standard
deviation (σ) of all cross-peak volumes per temperature.
𝜀95 =𝜎
√𝑛1.96
The number of integrated cross -peaks was typically six (n = 6), since one EXSY
spectrum delivers two cross-peaks on either side of the diagonal, and three different
mixing times were measured.
The obtained values for kex for each compound in D2O and CDCl3, and at different
temperatures, can be found in Tables 5 and 6. Three trends can be seen in the results.
Firstly, kex increases with increasing temperature. This is logical, since increasing the
temperature, increases the amplitude of atomic displacements, facilitating the rotation
around the amide bond. Secondly, kex increases with increasing amount of fluorine
atoms in the molecule. As already discussed in Chapter 1, the fluorine atoms withdraw
electron density from the amide bond, decreasing its double bond character and thus
making rotation around the bond easier. Lastly, a clear solvent effect is visible.
Exchange in D2O is strikingly slower than in CDCl3, and in D2O a difference in kex can
be observed between the monofluorinated molecules, while they have approximately
the same kex values in CDCl3, so the solvent effect is not uniform. There is a change
in charge separation in the amide bond during the rotation. This causes the carbonyl
oxygen atom to have a larger partial negative charge in a planar amide (equilibrium
state) than in an orthogonal amide (transition state). Therefore, the equilibrium state is
more polar than the transition state. This explains why rotation over the amide bond is
slower in D2O, which shows better solvation for polar compounds than CDCl3 [47].
56
Table 5 kex values in s-1 for the CDCl3 samples at different temperatures. Determined via EXSY: all temperatures for Ac-Pro-OMe and 303 K for the other samples. All others determined via selective inversion recovery.
T(K) Ac-Pro-OMe Ac-(4R)-FPro-OMe Ac-(4S)-FPro-OMe Ac-4,4-F2Pro-OMe
298 0.0973 ± 0.0073 0.2479 ± 0.0087 0.2058 ± 0.0205 0.7372 ± 0.0465
303 0.1849 ± 0.0181 0.4239 ± 0.0348 0.4327 ± 0.0105 1.4768 ± 0.0113
308 0.3295 ± 0.0389 0.7309 ± 0.0098 0.7544 ± 0.0125 2.5056 ± 0.0658
Table 6 kex values in s-1 for the D2O samples at different temperatures. Determined via EXSY: all temperatures for Ac-Pro-OMe and 308 K for the other samples. All others determined via selective inversion recovery.
T(K) Ac-Pro-OMe Ac-(4R)-FPro-OMe Ac-(4S)-FPro-OMe Ac-4,4-F2Pro-OMe
308 0.0354 ± 0.0080 0.1069 ± 0.0105 0.0415 ± 0.0063 0.1481 ± 0.0094
333 0.3693 ± 0.0086 0.8526 ± 0.0441 0.5200 ± 0.0269 1.7008 ± 0.0402
The result for Ac-Pro-OMe at 308 K can be compared with the result obtained by Eberhardt et al. [48]. They found a value for kex of 0.024 ± 0.17 s-1. Given the large uncertainty on their result, both results are comparable.
Obtained values for Ea; A; ΔH‡; and ΔS‡ can be found in Table 7. These values were
only obtained for CDCl3 samples, since in D2O, only two data points were collected
due to time restrictions, which is insufficient to perform linear regression. The first
observation that can be made is that the activation energy and activation enthalpy for
the backward reaction (from trans to cis) are always higher than for the forward
reaction. This is logical since the trans conformer is more stable than the cis conformer,
so the activation energy or activation enthalpy for the backward reaction will be the
sum of the activation energy of the forward reaction and the energy difference between
the cis and the trans conformer. Thus, according to the K values determined in the
previous section, the difference in Ea between the forward and the backward reaction
should be the smallest for Ac-(4S)-FPro-OMe and the largest for Ac-(4R)-FPro-OMe
with the other two molecules in between. The difference is not the largest for Ac-(4R)-
FPro-OMe, but keep in mind that these values are only determined by a three point
linear regression, so the error is still quite large (see Table 7). It can be noticed that
ΔH‡ is always about 2.5 kJ/mol less than Ea, a trend that is also visible in the results of
Thomas et al. [49]. ΔH‡ and ΔS‡ were also determined by Renner et al. [4], and when
comparing the results, it can be noticed that they are not in agreement with each other,
and sometimes even opposing. However, it needs to be noticed that the samples
studied by Renner et al. were dissolved in D2O, while our studied samples were
dissolved in CDCl3, and solvent can surely have an effect on the difference in stability
of the cis/trans forms (vide supra). It also needs to be emphasized that the results
obtained here are only preliminary and have high uncertainties due to the limited
number of data points. However, it is striking that the values determined here are
57
following a trend that correlates with the puckering equilibrium, as Ac-(4R)-FPro-OMe
has the highest activation enthalpy and the lowest activation entropy, while Ac-(4S)-
FPro-OMe has the lowest activation enthalpy and the highest activation entropy, and
both the activation enthalpy and activation entropy of Ac-Pro-OMe and Ac-4,4-F2Pro-
OMe have intermediate values. It could be that n → π* interactions also play a role in
the transition state, and thus stabilize and order it, just as was proposed before when
discussing the thermodynamic data (vide supra). But still, one has to be cautious in
interpreting the results, and more data points need to be measured to reduce the
uncertainties of these results.
Table 7 Ea values in kJ/mol; A values in 1015 s-1; ΔH‡ values in kJ/mol; and ΔS‡ values in J/mol K for the CDCl3 samples, for both the forward and backward reaction.
Ac-Pro-OMe Ac-(4R)-FPro-OMe Ac-(4S)-FPro-OMe Ac-4,4-F2Pro-OMe
k1 k-1 k1 k-1 k1 k-1 k1 k-1
Ea 92.5 ± 11.6 95.5 ± 14.8 82.0 ± 0.55 84.5 ± 14.4 98.9 ± 41.7 99.7 ± 53.5 92.6 ± 41.6 96.0 ± 37.9
A 1.26 ± 5.83 1.11 ± 6.49 0.05 ± 0.10 0.03 ± 0.18 28.6 ± 473 23.7 ± 504 9.51 ± 157 12.5 ± 188
ΔH‡ 89.9 ± 11.7 93.0 ± 14.8 79.5 ± 0.54 82.0 ± 14.3 96.4 ± 41.7 97.2 ± 53.5 90.0 ± 41.6 93.5 ± 37.9
ΔS‡ 35.7 ± 38.7 34.7 ± 49.0 8.41 ± 18.0 4.96 ± 47.3 61.7 ± 138 60.1 ± 177 52.5 ± 137 54.8 ± 125
58
5. General conclusion
5.1. Concluding remarks
The goal of this thesis was to evaluate the potential of the recently developed NMR
methodology (PSYCHEDELIC [31]) for measuring 1H-1H and 1H-19F couplings in
available fluorinated proline variants, and to use these couplings in combination with
generalized Karplus equations [39, 42] and the Altona-Sundaralingam formalism [35]
to reinvestigate their ring pucker. Although the pucker of the studied compounds has
previously been established [1], an extensive analysis of all 3JHH and 3JHF couplings
using the aforementioned conformational analysis strategy has not yet been
performed. Also the thermodynamics and kinetics of the cis/trans equilibrium of the
amide bond of these (fluoro)proline analogues were investigated. The explored
methodology, and identification of potential pitfalls, will also serve to swiftly engage into
the study of novel fluorinated proline compounds whose properties are fully unknown,
such as the 3,4-difluorinated prolines.
Generally, PSYCHEDELIC provides a very straightforward way for obtaining
couplings, as the 2D J-resolved spectrum that is generated is easy to interpret.
However, one major problem still plagues unhindered interpretation of PSYCHEDELIC
spectra and pure shift spectra in general, and that is when protons are strongly
coupled. How to accurately extract individual couplings from proton signals that are
involved in strong coupling, still remains a challenge for these types of experiments.
This caused that not all individual couplings could be measured for the fluoroprolines.
It is clear that future developments in this research area will be about resolving strong
coupling artefacts.
Puckering analysis of Ac-Pro-OMe, Ac-(4R)-FPro-OMe and Ac-(4S)-FPro-OMe, all
confirmed the results from puckering analyses conducted in the past. Ac-Pro-OMe is
predominantly Cγ endo, but with a considerable fraction of Cγ exo (ca. 15% for the cis
form, and 30% for the trans form, which is logical since trans prefers the exo pucker).
Ac-(4R)-FPro-OMe and Ac-(4S)-FPro-OMe both show a strong preference for a single
pucker, being Cγ exo and Cγ endo respectively. The presence of a potential second,
minimally populated pucker could not be detected using the current state of the
methodology. The puckering analysis of Ac-4,4-F2Pro-OMe confirmed the assumption
made in literature [4] that there will be no strong preference for one pucker but rather
59
an equilibrium between multiple puckers. However, the assumption that Ac-4,4-F2Pro-
OMe will have a similar puckering behaviour as Ac-Pro-OMe could not be confirmed
using the available 1H-19F Karplus curves. Although, based on comparison of the only
two vicinal 1H-1H couplings alone, it is indeed not unreasonable to assume this. Thus,
it is clear that there are still some challenges to be overcome, especially when the use
of 1H-19F couplings is necessary.
The pucker analysis was supplemented with a thermodynamic and kinetic study of the
cis/trans equilibrium involving the amide bond. This delivered results that are in
agreement with observations in literature, and that could be rationalized. The trans
conformer is always more prevalent than the cis [1], and a correlation between the
puckering preference and the cis/trans equilibrium was demonstrated [5]. The trans
conformer was found to be more predominant in D2O than in CDCl3 [46]. The exchange
rate between the cis and trans forms increases with increasing number of fluorine
atoms [1], and exchange always occurs faster in CDCl3 than in D2O [47].
5.2. Future prospects
In terms of puckering analysis, a proper validation of the used 1H-19F Karplus relation
in the context of fluoroprolines is in order. Another issue is the lack of correct group
electronegativities and bond angles, which further affect the accuracy of the used
Karplus relations. Extensive computational studies of the investigated molecules can
possibly offer a combined solution for these problems.
In terms of kinetic studies, future prospects include extending the temperature
dependent data to increase the precision of the obtained Arrhenius or Eyring
parameters, and to confirm or disprove the rationalizations that were made concerning
the observed trends between the different molecules. Again, computational studies
may be able to point out which interactions are determining for these parameters.
In the future, it will be the intention to use these methodologies to perform puckering
analyses and thermodynamic and kinetic studies on the less well-studied or fully
unknown fluoroproline analogues, such as the 3-fluoroprolines and the 3,4-
difluoroprolines, and the NMe2 derivatives instead of the OMe derivatives, as these
more closely resemble proline residues incorporated in a peptide sequence [46]. Then,
these results can be used to create a library of possible fluoroproline analogues and
their respective conformational preferences. Eventually, the purpose is to incorporate
60
fluoroproline in proline rich regions of intrinsically disordered proteins to modulate their
properties or to facilitate their study using NMR. As already indicated in Chapter 1, the
potential of 19F NMR as a site specific structural analysis technique is anticipated to be
considerable, because of the absence of amide protons in PRR motifs. Gaining
knowledge of dynamics and structure of PRRs can possibly lead to a better
understanding of the protein-protein interactions that lead to molecular recognition,
signal transduction and gene regulation.
61
6. Supporting information
6.1. NMR experiments
6.1.1. The basics of NMR
NMR is a structural analysis technique based on nuclear magnetism, which results
from the fact that a nucleus has an intrinsic magnetic moment, related to its nuclear
spin. When a spin ½ nucleus is placed in an external magnetic field (B0), the magnetic
moment interacts with the magnetic field (Zeeman effect), leading to a splitting of
energy levels according to the spin state (labelled α and β). The nuclear spins will also
precess around B0 with an angular frequency ω0 = -γB0 (called the Larmor frequency),
with γ the gyromagnetic ratio of the nucleus. The ensemble of spins in an NMR sample
leads to a net macroscopic nuclear magnetization, aligned and proportional with the
external magnetic field. By means of radio frequency (RF) pulses, this magnetization
can be tilted orthogonal to the magnetic field, where it will precess with a frequency ν0
(= ω0/2π). This, in turn, induces an oscillating current with the same frequency ν0 in a
detection solenoid. Fourier transform of this oscillating signal from the time domain to
the frequency domain yields an NMR spectrum. Since the value of γ differs for each
isotope, the measured frequencies will be nucleus dependent, hence the different
types of NMR (1H NMR, 13C NMR, 19F NMR,…). The exact magnetic field experienced
by a nucleus also depends on the chemical environment of the nucleus, leading to a
distinctive frequency for each nucleus in a molecule. Because this frequency is
proportional to the magnitude of the external B0-field, the δ-value, a B0 independent
measure for the chemical shift is usually employed, reported in parts per million (ppm)
rather than Hertz (Hz), against a reference [50].
6.1.2. 2D 1H-1H COSY
In a COSY (COrrelation SpectroscopY) spectrum, which is a homonuclear 2D
spectrum (both dimensions represent the same type of nucleus, 1H in our case), cross-
peaks can be seen between two 1H’s between which there exists a scalar coupling
[51]. In most cases, a cross-peak indicates that the two 1H’s are positioned geminal or
vicinal to each other. However, when longer-range couplings are present, cross-peaks
can be observed between 1H’s separated by more than three bonds.
62
6.1.3. 2D 1H-13C HSQC
An HSQC (Heteronuclear Single Quantum Correlation) spectrum allows one to see
correlations between 1H and 13C resonances that are coupled through a one-bond 1JCH
coupling. This can facilitate the assignment of crowded parts of the 1H spectra, where
there is lot of overlap of peaks, as such peaks may be resolved along the 13C
dimension. This type of spectrum also offers a straightforward way to detect geminal
1H signals, since these should correlate with the same 13C chemical shift. In this
research, multiplicity edited HSQC (m.e. HSQC) spectra were used, which means that
the sign of the cross-peak encodes the 13C multiplicity. Positive signs indicate CH- or
CH3-groups, while negative signs correspond to CH2-groups [51].
6.1.4. 2D 1H-13C HMBC
In an HMBC (heteronuclear multiple bond correlation) spectrum, one retrieves
correlations between coupled 1H and 13C spins that are separated by more than one
bond. In practice, 1H and 13C spins separated by two or three bonds possess a coupling
strong enough to lead to a cross-peak, although longer-range couplings may
sometimes be present. The HMBC spectrum is particularly useful to assign quaternary
carbons, as these cannot be assigned from an HSQC.
6.1.5. 2D 1H-1H NOESY
In a NOESY (Nuclear Overhauser Enhancement SpectroscopY) spectrum, the
correlations that are visible, result from nuclear magnetic dipole-dipole interactions
between the spins. These interactions thus occur through-space, in contrast with the
previously described 2D experiments, where the correlations resulted from through-
bond scalar couplings. Under the right conditions, the intensity of an nOe cross-peak
is inversely proportional to the sixth power of the distance between the 1H’s involved.
This implies that only 1H’s that are spatially close, will exhibit a detectable nOe cross-
peak. The strength of an nOe cross-peak depends on the NOESY mixing time, which
is the time during which the nOe has to build up [51].
6.1.6. 2D 1H-19F HOESY
A HOESY (Heteronuclear Overhauser Enhancement SpectroscopY) spectrum is
essentially a heteronuclear version of a NOESY spectrum. In this project, 1H-19F
HOESY spectra were used in which the direct signal acquisition took place on 1H, in
63
other words, the 1H dimension was the direct dimension (F2), while the 19F dimension
was the indirect dimension (F1).
6.1.7. Experimental
Samples were prepared by dissolving 2.05 mg of Ac-Pro-OMe, 2.27 mg of Ac-(4R)-
FPro-OMe, 2.27 mg of Ac-(4S)-FPro-OMe, and 2.49 mg of Ac-4,4-F2Pro-OMe into ca.
600 µl of CDCl3 (Eurisotop, 99.8%) or D2O (Eurisotop, 99.9%).
All NMR measurements were performed on either one of four different spectrometers:
a Bruker Avance II spectrometer equipped with either a 5mm 1H-13C-31P TXI-Z probe
or a 5mm 1H-13C-19F TXO-Z probe, operating at 1H, 13C and 19F frequencies of
500.13 MHz, 125.76 MHz and 470.59 MHz respectively; a Bruker Avance III
spectrometer equipped with a 5mm BBI-Z probe, operating at 1H and 13C frequency of
500.13 MHz and 125.76 MHz respectively; a Bruker Avance II spectrometer equipped
with a 5mm 1H-13C-31P TXI-Z probe, operating at 1H and 13C frequency of 700.13 MHz
and 176.05 MHz respectively; or a Bruker Avance II spectrometer equipped with a
5mm 1H-BB-19F TBI-Z probe, operating at 1H and 19F frequency of 400.13 MHz and
376.50 MHz respectively. All measurements were performed at 298.0 K unless
mentioned otherwise.
For 1D 19F spectra (with or without 1H decoupling), a double spin echo of a few
microseconds long and a 16-step phase cycle was applied to suppress the 19F
background signal from the probe [52]. 2D spectra measured for spectral assignment
included 1H-1H gCOSY, 1H-1H NOESY using zero-quantum suppression [53] during
the 500 ms mixing time, 1H-13C multiplicity edited gHSQC using adiabatic 13C pulses,
a 1H-13C gHMBC optimized for couplings of 7 Hz, and a 1H-19F HOESY with a mixing
time of 500 ms. Spectral windows were optimized to ensure detection of all correlations
in every dimension. Typically, total time domain sampled in the F2 and F1 dimensions
were 2048 and 512 respectively. Prior to Fourier transform, all spectra were multiplied
with a square cosine bell window function (square sine bell in the case of the COSY
and HMBC) and zero filled until a 2048×2048 real data matrix was obtained in the
frequency dimension.
Pure shift spectra were recorded using the PSYCHE [30] method, available from the
University of Manchester NMR methodology group website
(http://nmr.chemistry.manchester.ac.uk/), and applying an interferogram acquisition
64
scheme [29]. The length of the time domain data chunks had a duration matched to
the dwell time in the t2 time domain dimension to avoid discontinuities between chunks.
The duration of the data chunks was chosen to be between 50 ms and 100 ms.
Typically, 32 data chunks or more were acquired. The pure shift data was
reconstructed using a macro available from the University of Manchester NMR
methodology group website. The RF-field strength of the amplitude-modulated double
chirp pulse (PSYCHE element) was typically set to achieve about a 20° flip angle.
PSYCHEDELIC [31] spectra were set up similarly as the PSYCHE pure shift spectra.
The spectral width of the 2DJ indirect dimension was set to ca. 20 Hz, and fine-tuned
to be exactly in a power of two ratio with the direct dimension spectral width in order to
facilitate 45° tilting. Resolution enhancing Lorenz-to-Gauss window functions were
applied, width line broadening parameters (lb) and Gaussian broadening (gb)
parameters in TopSpin were standardly set to –1.5 and 0.3 respectively along both
dimensions of the 2DJ spectrum, or as a function of the required resolution to measure
couplings. For the 1H selective pulses, 180° RSNOB [54] or RE-BURP [55] pulses were
used, with a duration chosen as a function of the spectral region to be excited. For 1H-
19F couplings, an in-house pulse sequence was used that work similarly as
PSYCHEDELIC, but applies 19F selective pulses.
For 19F T1 measurements, inversion recovery experiments with power gated 1H
decoupling were used, with an acquisition time of 3.5 s, and an interscan delay of
12.5 s. Typically, 12 points were sampled, with recovery delays ranging from 10 µs to
18 s seconds. For the selective inversion recovery experiments, 16 data points were
sampled up to a 20 s recovery delay, and a 16 s interscan delay was used. For 1H T1
measurements, regular inversion recovery experiments with similar delays as for 19F
were used. For 2D 1H-1H EXSY experiments, a similar experimental set up and
spectrum processing of the 2D NOESY spectra was used. An interscan delay of 15 s
was used. The spectral width was reduced in order to limit the number of time domain
points needed in t1 to achieve sufficient resolution in order to resolve the methyl cross-
peaks from the diagonal peaks. A similar set up was applied for the 19F-19F EXSY
experiments, except that no zero-quantum suppression was applied.
65
6.2. 1H chemical shift tables
All spectra were recorded at 298 K at a 700 MHz (Ac-Pro-OMe and Ac-(4S)-FPro-OMe
in CDCl3) or 500 MHz (rest) spectrometer. The reported chemical shift values are in
ppm. Which label refers to which 1H can be seen in the figure below.
Figure 35 1H labels.
This type of labelling scheme is more convenient than working with pro R and pro S as
it allows a more straightforward way of comparing the results. Suffix 1 indicates that
the hydrogen atom is pointing up, while suffix 2 indicates that the hydrogen atom is
pointing down. In most of the cases, the pro R hydrogen atom will be the one pointing
up, and the pro S hydrogen atom will be the one pointing down. For the β position in
Ac-Pro-OMe however, the opposite is true. In case of the difluorinated analogue, no γ
chemical shifts can be reported as they are both substituted by fluorine. For the
monofluorinated analogues, only one of the γ hydrogen atoms is replaced by fluorine,
a chemical shift value for the one that is not substituted will be reported.
6.2.1. CDCl3 samples
Table 8 1H chemical shift values for all CDCl3 samples.
Ac-Pro-OMe Ac-(4R)-FPro-OMe Ac-(4S)-FPro-OMe Ac-4,4-F2Pro-OMe
trans cis trans cis trans cis trans cis
Me-[CO] 2.12 2.01 2.13 2.02 2.15 2.08 2.12 2.03
Me-[O] 3.76 3.80 3.78 3.82 3.77 3.82 3.79 3.84
α 4.52 4.41 4.61 4.60 4.81 4.53 4.74 4.62
β1 2.21 2.31 2.63 2.78 2.37 2.44 2.72 2.84
β2 2.02 2.19 2.15 2.31 2.56 2.75 2.51 2.72
γ1 2.02 1.95 / / 5.33 5.27 / /
γ2 2.10 1.95 5.33 5.26 / / / /
δ1 3.53 3.59 3.84 4.27 3.80 3.80 3.91 4.05
δ2 3.68 3.66 3.84 3.56 3.93 3.91 4.00 3.84
66
6.2.2. D2O samples
Table 9 1H chemical shift values for all D2O samples.
Ac-Pro-OMe Ac-(4R)-FPro-OMe Ac-(4S)-FPro-OMe Ac-4,4-F2Pro-OMe
trans cis trans cis trans cis trans cis
Me-[CO] 2.04 1.93 2.07 1.96 2.08 2.00 2.05 1.97
Me-[O] 3.68 3.73 3.70 3.75 3.71 3.74 3.71 3.76
α 4.37 4.63 4.51 4.84 4.67 4.84 4.69 4.97
β1 2.22 2.28 2.64 2.77 2.47 2.51 2.81 2.90
β2 1.94 2.14 2.15 2.36 2.47 2.64 2.56 2.75
γ1 1.94 1.91 / / 5.36 5.32 / /
γ2 1.94 1.78 5.37 5.30 / / / /
δ1 3.57/3.59 3.40 3.95 4.05 3.83 3.67 4.03 3.88
δ2 3.57/3.59 3.47 3.83 3.46 3.92 3.71 4.03 3.77 It was not possible to assess pro R and pro S for the δ hydrogens of the trans conformer of Ac-Pro-OMe.
6.3. 13C chemical shift tables
All spectra were recorded at 298 K at a 700 MHz (Ac-Pro-OMe and Ac-(4S)-FPro-OMe
in CDCl3) or 500 MHz (rest) spectrometer. The reported chemical shift values are in
ppm. Which label refers to which 13C can be seen in the figure below.
Figure 36 13C labels.
6.3.1. CDCl3 samples
Table 10 13C chemical shift values for all CDCl3 samples.
Ac-Pro-OMe Ac-(4R)-FPro-OMe Ac-(4S)-FPro-OMe Ac-4,4-F2Pro-OMe
trans cis trans cis trans cis trans cis
Me-[CO] 22.27 22.23 22.25 21.69 22.24 22.08 22.05 21.00
Me-[O] 52.22 52.57 52.41 52.81 52.47 52.81 52.70 53.14
α 58.46 60.17 57.18 58.22 56.92 58.70 56.21 58.03
β 29.41 31.46 36.08 38.24 36.08 38.20 37.53 39.45
γ 24.78 22.80 91.87 90.21 92.15 90.78 126.22 *
δ 47.72 46.28 54.17 52.70 54.07 53.11 54.36 52.90
CO-Am 169.35 169.50 169.33 169.86 169.33 169.77 169.24 169.64
CO-Es 172.83 172.64 172.43 172.28 171.13 171.30 170.68 170.42 *γ 13C chemical shift of the Ac-4,4-F2Pro-OMe cis conformer could not be determined, correlations were not
visible in the HMBC.
67
6.3.2. D2O samples
Table 11 13C chemical shift values for all D2O samples.
Ac-Pro-OMe Ac-(4R)-FPro-OMe Ac-(4S)-FPro-OMe Ac-4,4-F2Pro-OMe
trans cis trans cis trans cis trans cis
Me-[CO] 21.16 21.16 21.25 20.74 21.23 21.12 21.12 20.12
Me-[O] 52.87 53.18 53.02 53.38 53.05 53.34 53.28 53.63
α 59.03 60.66 57.65 58.46 57.37 59.09 56.61 58.49
β 29.15 30.56 35.55 37.04 35.49 37.15 36.42 38.04
γ 24.20 22.29 92.75 91.32 93.37 92.24 126.13 126.02
δ 48.38 46.58 54.46 52.75 54.52 53.16 53.83 52.55
CO-Am 172.94 173.28 173.20 173.93 173.36 173.84 173.25 173.81
CO-Es 175.00 174.59 174.23 173.93 173.75 173.96 172.68 172.60
6.4. 19F chemical shift tables
All spectra were recorded at 298 K at a 400 MHz (Ac-4,4-F2Pro-OMe in D2O) or 500
MHz (rest) spectrometer. The reported chemical shift values are in ppm. Which label
refers to which 19F can be seen in the figure below. For Ac-(4R)-FPro-OMe, there is no
fluorine atom at the γ2 position, while for Ac-(4S)-FPro-OMe, there is no fluorine atom
at the γ1 position. In case of Ac-4,4-F2Pro-OMe, the fluorine at the γ1 position is pro
R, and the fluorine at the γ2 position is pro S.
Figure 37 19F labels.
6.4.1. CDCl3 samples
Table 12 19F chemical shift values for all CDCl3 samples.
Ac-(4R)-FPro-OMe Ac-(4S)-FPro-OMe Ac-4,4-F2Pro-OMe
trans cis trans cis trans cis
γ1 -176.96 -177.87 / / -98.19 -101.43
γ2 / / -172.85 -173.12 -99.11 -96.74
68
6.4.2. D2O samples
Table 13 19F chemical shift values for all D2O samples.
Ac-(4R)-FPro-OMe Ac-(4S)-FPro-OMe Ac-4,4-F2Pro-OMe
trans cis trans cis trans cis
γ1 -177.85 -177.79 / / -102.23 -104.71
γ2 / / -173.26 -173.25 -98.35 -95.46
6.5. 1H-1H coupling tables
All spectra were recorded at 298 K at a 700 MHz (Ac-(4R)-FPro-OMe in CDCl3) or 500
MHz (rest) spectrometer. The reported coupling constants are in Hz. The same labels
as in part 1 are used.
6.5.1. CDCl3 samples
Table 14 Measured 1H-1H couplings for all CDCl3 samples.
Ac-Pro-OMe Ac-(4R)-FPro-OMe Ac-(4S)-FPro-OMe Ac-4,4-F2Pro-OMe
trans cis trans cis trans cis trans cis
α - β1 8.7 8.6 8.0 8.3 9.9 9.5 9.3 9.4
α - β2 4.0 2.8 9.1 8.1 1.6° 0.8°° 5.4 3.2
β1 - γ1 7.6 9.0† / / 4.1 3.8 / /
β1 - γ2 8.6 9.0† 1.7° 1.7 / / / /
β2 - γ1 * 5.3† / / 1.2° 0.9°° / /
β2 - γ2 6.5 5.3† 4.3 4.4 / / / /
γ1 - δ1 7.0 8.1† / / 4.4 4.3 / /
γ1 - δ2 4.8 6.2† / / <1 <1 / /
γ2 - δ1 7.3 8.1† 2.2‡ 1.0° / / / /
γ2 - δ2 7.8 6.2† 2.2‡ 3.6 / / / / *: coupling could not be measured, †: measured coupling is an average of two couplings, ‡: measured coupling is
a sum of two couplings, °: coupling was measured with Lorentzian line broadening set to -3, °°: coupling was
measured with Lorentzian line broadening set to -5.
69
6.5.2. D2O samples
Table 15 Measured 1H-1H couplings for all D2O samples.
Ac-Pro-OMe Ac-(4R)-FPro-OMe Ac-(4S)-FPro-OMe Ac-4,4-F2Pro-OMe
trans cis trans cis trans cis trans cis
α - β1 8.8 8.9 7.8 8.7 2.9† 9.7 9.8 9.8
α - β2 4.7 2.6 10.1 8.2 2.9† 0.8°° 4.6 2.1
β1 - γ1 * 7.1 / / 1.6† 3.4 / /
β1 - γ2 * 11.1 1.3° 1.5° / / / /
β2 - γ1 * 3.5 / / 1.6† 1.1°° / /
β2 - γ2 * 6.9 3.8 4.3 / / / /
γ1 - δ1 * 7.6 / / 3.7 3.8 / /
γ1 - δ2 * 3.6 / / <1 <1 / /
γ2 - δ1 * 9.2 <1 0.8°° / / / /
γ2 - δ2 * 8.6 3.2 3.4 / / / / *: coupling could not be measured, †: measured coupling is an average of two couplings, ‡: measured coupling is
a sum of two couplings, °: coupling was measured with Lorentzian line broadening set to -3, °°: coupling was
measured with Lorentzian line broadening set to -5.
6.6. 1H-19F coupling tables
All spectra were recorded at 298 K at a 400 MHz (Ac-4,4-F2Pro-OMe in D2O) or 500
MHz (rest) spectrometer. The reported coupling constants are in Hz. The same labels
as indicated in the figure below are used.
Figure 38 Labels used for reporting 1H-19F couplings.
70
6.6.1. CDCl3 samples
Table 16 Measured 1H-19F couplings for all CDCl3 samples.
Ac-(4R)-FPro-OMe Ac-(4S)-FPro-OMe Ac-4,4-F2Pro-OMe
trans cis trans cis trans cis
β1 - F1 19.1 20.1 / / 13.8 11.8
β1 - F2 / / 40.5 42.2 14.2 20.0
β2 - F1 38.1 36.2 / / 13.7 7.4
β2 - F2 / / 18.8 15.9 13.7 *
δ1 - F1 * 21.3 / / 11.5 12.2
δ1 - F2 / / 33.6 37.5° 10.9 15.6
δ2 - F1 * 35.9 / / 13.3 9.9
δ2 - F2 / / 25.6 27.2 13.1 13.5 *: coupling could not be measured, †: measured coupling is an average of two couplings, °: coupling was
measured in the HSQC.
6.6.2. D2O samples
Table 17 Measured 1H-19F couplings for all D2O samples.
Ac-(4R)-FPro-OMe Ac-(4S)-FPro-OMe Ac-4,4-F2Pro-OMe
trans cis trans cis trans cis
β1 - F1 18.5 21.1 / / 11.8 10.3
β1 - F2 / / * 46.1° 19.6 21.5
β2 - F1 41.9 39.2 / / 10.6 4.2
β2 - F2 / / * 16.7 13.6 14.7
δ1 - F1 21.9 21.0 / / 10.9† 10.5
δ1 - F2 / / 38.7 * 14.5† 19.8
δ2 - F1 38.2 37.2 / / 10.9† 5.9
δ2 - F2 / / 25.0 * 14.5† 16.9 *: coupling could not be measured, †: measured coupling is an average of two couplings, °: coupling was
measured in the HSQC.
6.7. Puckering analysis
6.7.1. Advantages and disadvantages of the GUI
The advantages of the GUI are that it is user friendly, and data input is limited and
straightforward, certainly for the group electronegativities for which an editor can be
used. In the editor, one can graphically assign a group electronegativity value to every
possible substituent, and for a certain coupling, the program automatically picks the
needed λ’s and also automatically assigns to the correct positions in the Diez-Donders
equation. The program also allows the use of sums of couplings as input, which is
useful when no individual couplings could be measured but average couplings could
be obtained.
71
Figure 39 Input window of the graphical user interface.
However, there are some shortcomings. The program works as a black box, with
limited information on what happens on the background. It also only works with 1H-1H
couplings, which is problematic for Ac-4,4-F2Pro-OMe, for which only two 1H-1H
couplings can be used, which makes fitting of even one conformation impossible. For
example, for the monofluorinated molecules, a maximum of six experimental 1H-1H
couplings are available, which means two conformations can in principle be fitted,
though the results may be unreliable (using six experimental parameters to fit five
parameters). It is thus clear that 1H-19F couplings are essential for these compounds.
Furthermore, the above-described Barfield correction is not implemented in the GUI.
Finally, the generated pseudorotation plots for two conformations turned out not
straightforward to interpret, as were the uncertainties of the fitted parameters. In order
to overcome these problems, we resorted to writing a new fitting script that better suited
our needs.
For one case, a comparison of the generated plots by both the GUI and the MATLAB
script will be made (see Figure 6). In the GUI, a plot of the RMSD of every point in
conformational space is made, while in the MATLAB script every fitted conformation
during the Monte Carlo analysis is plotted. This yields two different plots, in case of the
GUI, a shallow RMSD well indicates a lower goodness-of-fit, while in case of the
72
MATLAB script, a broad spread of points indicates a higher uncertainty on the fitted
value due to measurement errors.
Figure 40 Pseudorotation plots for the cis conformer of Ac-(4S)-FPro-OMe in D2O. Left: From the MATLAB script. Right: From the GUI. Note that the polar coordinate is arranged differently.
6.7.2. Precision
Most 1H-1H couplings could be measured multiple times. One can select 1H A and
observe 1H B, but also the other way around. Furthermore, for fluorinated compounds,
the β, γ and δ protons are also splitted along F2 due to 1H-19F coupling, meaning that
the same coupling can be measured twice in the same PSYCHEDELIC spectrum.
Since all these splittings, though measured on different occasions, should yield the
same value, we are able to make an estimate of the precision of the measurement. We
found the spread in measured values to be in the order of magnitude of 0.1 Hz.
6.7.3. Input parameters
For the couplings, the same labels are used as in sections 6.4 and 6.5, only now F is
replaced by γ for the 1H-19F couplings. The λ value of H is 0, that of COOR is 0.42, that
of NR2 is 1.12, and that of F is 1.37. The λ values of CHFR and CF2R were
approximated by that of CH2F (0.65), those of CH2CH2NR2 and CH2CH(NR2)COOR
were approximated by that of CH2CH2OR (0.67), that of CH(NR2)COOR was
approximated by that of CH2COOR (0.68), and that of CH2NR2 was approximated by
that of CH2OR (0.68). All used λ values can be found in Table 11. The used values for
B, Tb and Pb (if applicable) can be found in Table 12. For parameter A, a value of 1
was always used. In Table 13, the used bond angles in the 1H-19F Karplus relation can
73
be found. Those for Ac-(4R)-FPro-OMe were obtained via the available crystal
structure, the values for the other molecules were obtained via computational
calculations.
Table 18 All used group electronegativities, both for the Diez-Donders and the H-C-C-F Karplus equation.
Ac-Pro-OMe Ac-(4R)-FPro-OMe Ac-(4S)-FPro-OMe Ac-4,4-F2Pro-OMe
λ1 λ2 λ3 λ4 λ1 λ2 λ3 λ4 λ1 λ2 λ3 λ4 λ1 λ2 λ3 λ4
α-β1 0 0.67 1.12 0.42 0 0.65 1.12 0.42 0 0.65 1.12 0.42 0 0.65 1.12 0.42
α-β2 0.67 0 1.12 0.42 0.65 0 1.12 0.42 0.65 0 1.12 0.42 0.65 0 1.12 0.42
β1-γ1 0 0.68 0.67 0 0.68 0 0 0.68 0.68 0 1.37 0.68 1.37 0.68 0.68 0
β1-γ2 0.68 0 0.67 0 0.68 0 0.68 1.37 0.68 0 0.68 0 0.68 1.37 0.68 0
β2-γ1 0 0.68 0 0.67 0 0.68 0 0.68 0 0.68 1.37 0.68 1.37 0.68 0 0.68
β2-γ2 0.68 0 0 0.67 0 0.68 0.68 1.37 0 0.68 0.68 0 0.68 1.37 0 0.68
γ1-δ1 0.67 0 0 1.12 0.67 0 0 1.12 0.67 1.37 0 1.12 0.67 1.37 0 1.12
γ1-δ2 0.67 0 1.12 0 0.67 0 1.12 0 0.67 1.37 1.12 0 0.67 1.37 1.12 0
γ2-δ1 0 0.67 0 1.12 1.37 0.67 0 1.12 0 0.67 0 1.12 1.37 0.67 0 1.12
γ2-δ2 0 0.67 1.12 0 1.37 0.67 1.12 0 0 0.67 1.12 0 1.37 0.67 1.12 0
Table 19 Values for B (deg), Tb (Hz) and Pb (deg) parameters. Tb and Pb are not applicable (n.a.) for transoid couplings (B ≠ 0).
B Tb Pb
α-β1 0 2 234
α-β2 -120 n.a. n.a.
β1-γ1 0 1 90
β1-γ2 -120 n.a. n.a.
β2-γ1 120 n.a. n.a.
β2-γ2 0 1 270
γ1-δ1 0 2 306
γ1-δ2 -120 n.a. n.a.
γ2-δ1 120 n.a. n.a.
γ2-δ2 0 2 126
74
Table 20 Used aFCC and aHCC angles (in degrees) as input for the 1H-19F Karplus relation. For Ac-(4S)-FPro-OMe and Ac-4,4-F2Pro-OMe, these were calculated for trans and cis separately.
Ac-(4R)-FPro-OMe trans
Ac-(4S)-FPro-OMe cis
Ac-(4S)-FPro-OMe trans
Ac-4,4-F2Pro-OMe cis
Ac-4,4-F2Pro-OMe
aFCC aHCC aFCC aHCC aFCC aHCC aFCC aHCC aFCC aHCC
β1 - F1 108.03 110.2 / / / / 112.52 108.52 112.42 108.38
β1 - F2 / / 108.04 109.68 108.47 109.7 109.27 108.52 109.89 108.38
β2 - F1 108.03 110.3 / / / / 112.52 110.92 112.42 110.94
β2 - F2 / / 108.04 111.07 108.47 110.89 109.27 110.92 109.89 110.94
δ1 - F1 107.72 110.5 / / / / 112.37 110.4 111.83 109.57
δ1 - F2 / / 109.59 111.95 109.57 111.19 110.42 110.4 110.39 109.57
δ2 - F1 107.72 109.3 / / / / 112.37 110.77 111.83 109.48
δ2 - F2 / / 109.59 110.45 109.57 109.13 110.42 110.77 110.39 109.48
6.7.4. Output parameters
All fitted parameters and RMSD values from the puckering analysis can be found in
Tables 14, 15, 16 and 17 (one molecule per table). The residuals (the difference
between the calculated and experimental couplings) can be found in Tables 18, 19,
20 and 21 (one molecule per table). The residuals are defined as follows:
𝐽𝑒𝑥𝑝𝑡 − 𝐽𝑐𝑎𝑙𝑐
Table 21 Puckering fit output for Ac-Pro-OMe.
CDCl3 D2O
trans cis trans cis
1 conformer
P 76.4° ± 64.3° 175.7° ± 7.9° / 300.7° ± 53.6°
νmax 0.0° ± 0.7° 17.3° ± 1.3° / 0.0° ± 0.6°
RMSD 2.56 Hz 0.63 Hz / 3.18 Hz
2 conformers
P1 17.1° ± 7.1° 111.5° ± 19.8° / 13.4° ± 10.2°
νmax,1 49.7° ± 17.5° 143.2° ± 99.2° / 43.7° ± 20.5°
P2 187.9° ± 5.9° 185.0° ± 52.0° / 182.8° ± 3.7°
νmax,2 31.5° ± 7.5° 17.9° ± 33.7° / 36.4° ± 4.4°
x1 29.5% ± 6.2% 7.3% ± 50.5% / 16.9% ± 4.8%
RMSD 0.54 Hz 0.51 Hz / 0.81 Hz
2 conformers with Barfield
P1 19.3° ± 5.0° / / 17.7° ± 10.7°
νmax,1 47.7° ± 14.0° / / 39.0° ± 20.3°
P2 183.9° ± 5.0° / / 181.2° ± 3.6°
νmax,2 31.4° ± 6.3° / / 36.5° ± 4.8°
x1 29.6% ± 5.8% / / 17.6% ± 6.3%
RMSD 0.73 Hz / / 1.00 Hz
75
Table 22 Puckering fit output for Ac-(4R)-FPro-OMe.
CDCl3 D2O
trans cis trans cis
1 conformer without 1H-19F
P 24.9° ± 2.1° 24.1° ± 1.7° 13.5° ± 1.7° 25.0° ± 1.7°
νmax 48.0° ± 2.0° 40.7° ± 1.1° 43.5° ± 0.9° 41.1° ± 1.2°
RMSD 0.94 Hz 0.37 Hz 0.44 Hz 0.45 Hz
1 conformer with 1H-19F
P 24.0° ± 1.8° 13.1° ± 0.7° 3.1° ± 0.7° 14.5° ± 0.8°
νmax 45.9° ± 0.8° 38.6° ± 0.2° 44.5° ± 0.3° 40.3° ± 0.2°
RMSD 1.44 Hz 1.69 Hz 0.68 Hz 0.98 Hz
2 conformers with 1H-19F
P1 108.2° ± 39.8° 219.8° ± 3.2° 16.2° ± 63.3° 54.4° ± 2.7°
νmax,1 184.7° ± 77.7° 104.0° ± 3.4° 44.4° ± 98.7° 147.2° ± 7.3°
P2 23.9° ± 4.1° 8.0° ± 1.2° 59.0° ± 46.4° 22.0° ± 1.2°
νmax,2 47.1° ± 8.9° 37.5° ± 0.5° 150.0° ± 92.0° 41.2° ± 0.5°
x1 9.2% ± 21.0% 15.3% ± 1.0% 96.0% ± 85.9% 8.0% ± 0.6%
RMSD 0.83 Hz 0.49 Hz 0.52 Hz 0.41 Hz
1 conformer without 1H-19F with Barfield
P 24.9° ± 2.2° 205.7° ± 0.5° 13.5° ± 1.8° 25.0° ± 1.7°
νmax 48.0° ± 2.1° 142.5° ± 0.8° 43.5° ± 1.0° 41.1° ± 1.2°
RMSD 0.94 Hz 3.67 Hz 0.44 Hz 0.45 Hz
Table 23 Puckering fit output for Ac-(4S)-FPro-OMe.
CDCl3 D2O
trans cis trans cis
1 conformer without 1H-19F
P 206.5° ± 2.9° 199.3° ± 3.0° 165.2° ± 5.4° 203.2° ± 2.6°
νmax 37.1° ± 1.1° 38.0° ± 1.1° 56.3° ± 6.8° 41.6° ± 1.2°
RMSD 0.74 Hz 0.68 Hz 1.02 Hz 0.76 Hz
1 conformer with 1H-19F
P 179.4° ± 1.0° 175.8° ± 0.8° 166.8° ± 9.8° 198.7° ± 2.1°
νmax 40.7° ± 0.3° 45.9° ± 0.3° 56.4° ± 15.7° 50.7° ± 0.7°
RMSD 1.90 Hz 2.46 Hz 3.56 Hz 1.51 Hz
2 conformers with 1H-19F
P1 135.3° ± 4.6° 175.8° ± 64.8° 158.5° ± 31.7° 190.6° ± 49.0°
νmax,1 150.0° ± 8.5° 45.9° ± 74.8° 67.6° ± 64.4° 48.3° ± 54.5°
P2 180.6° ± 11.9° 175.8° ± 50.5° 166.8° ± 72.5° 198.7° ± 52.7°
νmax,2 40.7° ± 73.7° 45.9° ± 59.1° 56.4° ± 25.2° 50.7° ± 67.6°
x1 7.5% ± 7.7% 49.6% ± 83.1% 0.0% ± 78.5% 0.0% ± 87.3%
RMSD 1.81 Hz 2.46 Hz 3.56 Hz 1.51 Hz
1 conformer without 1H-19F with Barfield
P 187.2° ± 4.4° 185.4° ± 4.6° / 193.7° ± 4.1°
νmax 31.7° ± 1.1° 34.3° ± 1.1° / 38.1° ± 1.0°
RMSD 1.24 Hz 1.11 Hz / 1.25 Hz
76
Table 24 Puckering fit output for Ac-4,4-F2Pro-OMe.
CDCl3 D2O
trans cis trans cis
1 conformer
P 144.6° ± 98.4° 167.7° ± 1.7° 133.9° ± 1.3° 188.1° ± 1.3°
νmax 0.0° ± 0.4° 21.3° ± 0.4° 31.3° ± 0.8° 29.4° ± 0.2°
RMSD 5.03 Hz 4.06 Hz 4.63 Hz 2.73 Hz
2 conformers
P1 40.4° ± 0.9° 223.1° ± 1.3° 50.5° ± 1.1° 47.4° ± 2.3°
νmax,1 116.2° ± 1.8° 111.8° ± 2.0° 169.3° ± 5.1° 103.4° ± 2.5°
P2 84.3° ± 2.2° 247.5° ± 3.5° 102.8° ± 15.2° 179.1° ± 1.8°
νmax,2 23.9° ± 1.9° 17.8° ± 1.9° 13.2° ± 5.8° 29.2° ± 0.6°
x1 40.4% ± 0.6% 41.0% ± 0.7% 53.8% ± 1.4% 19.0% ± 0.6%
RMSD 1.12 Hz 0.76 Hz 0.47 Hz 1.29 Hz
2 conformers with Barfield
P1 40.4° ± 0.9° 38.1° ± 3.4° / 49.2° ± 2.4°
νmax,1 116.2° ± 1.9° 72.4° ± 10.1° / 104.7° ± 2.7°
P2 84.3° ± 2.3° 162.2° ± 5.7° / 175.7° ± 1.8°
νmax,2 23.9° ± 1.9° 30.9° ± 1.9° / 29.7° ± 0.7°
x1 40.4% ± 0.6% 28.4% ± 1.1% / 19.3% ± 0.6%
RMSD 1.12 Hz 1.08 Hz / 1.36 Hz
Table 25 Residuals of the puckering analysis of Ac-Pro-OMe, only for fitting two conformers, without Barfield correction.
CDCl3 D2O
trans cis cis
α-β1 1.0 0.3 1.4
α-β2 0.3 0.3 0.3
β1-γ1 0.4 0.8† 0.1
β1-γ2 0.7 0.8† 1.4
β2-γ1 * 0.2† 0.7
β2-γ2 -0.7 0.2† -0.1
γ1-δ1 0.0 0.5† 0.6
γ1-δ2 0.5 0.7† 0.9
γ2-δ1 -0.1 0.5† 0.4
γ2-δ2 0.5 0.7† 0.9 *: coupling was not measured and could therefore not be used. †: residual of a sum of couplings.
77
Table 26 Residuals of the puckering analysis of Ac-(4R)-FPro-OMe, only for fitting one conformer, without Barfield correction
without 1H-19F couplings with 1H-19F couplings
CDCl3 D2O CDCl3 D2O
trans cis trans cis trans cis trans cis
α-β1 0.5 0.5 0.9 0.9 0.5 1.0 1.6 1.5
α-β2 0.5 0.1 0.6 0.3 0.6 -0.9 -0.4 -0.8
β1-γ 0.1 0.7 -0.1 0.5 0.3 0.6 -0.3 0.3
β2-γ 0.6 -0.1 0.0 -0.2 0.4 0.0 0.2 0.0
γ-δ1 -1.9† 0.0 0.0 -0.3 -1.9† 0.2 0.1 -0.1
γ-δ2 -1.9† 0.3 -0.1 0.1 -1.9† -0.2 -0.4 -0.2
β1-F / / / / -1.6 -3.0 -0.7 -1.2
β2-F / / / / -2.7 -2.9 0.2 -0.5
δ1-F / / / / * -1.6 -0.1 -1.0
δ2-F / / / / * -2.4 -0.9 -2.0 *: coupling was not measured and could therefore not be used. †: residual of a sum of couplings.
Table 27 Residuals of the puckering analysis of Ac-(4S)-FPro-OMe, only for fitting one conformer, without Barfield correction.
without 1H-19F couplings with 1H-19F couplings
CDCl3 D2O CDCl3 D2O
trans cis trans cis trans cis trans cis
α-β1 1.4 1.3 -0.4† 1.4 2.7 3.0 -0.4† 2.1
α-β2 0.0 -0.4 -0.4† -0.4 0.9 0.0 -0.4† 0.0
β1-γ -0.9 -0.8 -2.0† -0.9 0.1 0.4 -2.0† 0.2
β2-γ 0.4 0.0 -2.0† 0.0 -0.1 -0.9 -2.0† -0.9
γ-δ1 0.7 0.6 -0.1 0.6 0.2 0.4 0.0 1.4
γ-δ2 0.1 0.2 0.2 -0.1 0.3 0.2 0.1 -0.7
β1-F / / / / -0.9 -0.2 * 3.2
β2-F / / / / -1.6 -1.5 * 0.5
δ1-F / / / / 3.3 5.1 6.9 *
δ2-F / / / / 3.6 4.6 4.9 * *: coupling was not measured and could therefore not be used. †: residual of a sum of couplings.
Table 28 Residuals of the puckering analysis of Ac-4,4-F2Pro-OMe, only for fitting two conformers, without Barfield correction.
CDCl3 D2O
trans cis trans cis
α-β1 1.2 0.8 1.1 1.7
α-β2 0.3 -0.2 0.2 0.0
β1-F1 0.7 -0.6 0.3 -2.0
β1-F2 0.6 -0.5 0.3 -1.3
β2-F1 0.3 -0.3 0.0 -1.9
β2-F2 0.0 * -0.3 -0.1
F1-δ1 1.1 0.9 -0.2† 0.6
F1-δ2 2.2 1.1 -0.2† 1.6
F2-δ1 1.8 1.3 -0.3† 1.1
F2-δ2 0.8 0.3 -0.3† 0.8 *: coupling was not measured and could therefore not be used. †: residual of a sum of couplings.
78
6.7.5. Pseudorotation plots
In the following figures, the ordering will always be as follows unless stated otherwise. Left: CDCl3 samples. Right: D2O samples. Top: trans conformers. Bottom: cis conformers.
6.7.5.1. Pseudorotation plots generated by the GUI
Ac-Pro-OMe
Figure 41 Pseudorotations plots from the GUI for Ac-Pro-OMe.
79
Ac-(4R)-FPro-OMe
Figure 42 Pseudorotation plots from the GUI for Ac-(4R)-FPro-OMe.
80
Ac-(4S)-FPro-OMe
Figure 43 Pseudorotation plots from the GUI for Ac-(4S)-FPro-OMe.
81
6.7.5.2. Pseudorotation plots generated by the MatLab script
Warning: not all pseudorotation plots depicted below have the same scale!
Ac-(4R)-FPro-OMe with 1H-19F couplings
Figure 44 Pseudorotation plots from the MatLab script for Ac-(4R)-FPro-OMe, with use of 1H-19F couplings. For fitting one conformer.
82
Figure 45 Pseudorotation plots from the MatLab script for Ac-(4R)-FPro-OMe, with use of 1H-19F couplings. For fitting two conformers.
83
Ac-(4S)-FPro-OMe with 1H-19F couplings
Figure 46 Pseudorotation plots from the MatLab script for Ac-(4S)-FPro-OMe, with use of 1H-19F couplings. For fitting one conformer.
84
Figure 47 Pseudorotation plots from the MatLab script for Ac-(4S)-FPro-OMe, with use of 1H-19F couplings. For fitting two conformers.
85
Fitting with the Barfield correction
Figure 48 Pseudorotation plots from the MatLab script for Ac-(4R)-FPro-OMe, with Barfield correction, without use of 1H-19F couplings. For fitting one conformer.
86
Figure 49 Pseudorotation plots from the MatLab script for Ac-(4S)-FPro-OMe, with Barfield correction, without use of 1H-19F couplings. For fitting one conformer.
87
Figure 50 Pseudorotation plots from the MatLab script for Ac-4,4-F2Pro-OMe, with Barfield correction, with use of 1H-19F couplings. For fitting two conformers.
88
Figure 51 Pseudorotation plots from the MatLab script for Ac-Pro-OMe, with Barfield correction. For fitting two conformers. Left: trans CDCl3. Right: cis D2O.
6.7.6. Fitting fractions Cγ exo/endo for Ac-4,4-F2Pro-OMe based on RMSD
The obtained RMSD plots for this procedure can be found here. 1H-19F couplings were used for this fitting.
Figure 52 trans Ac-4,4-F2Pro-OMe in CDCl3.
0
2
4
6
8
10
12
14
16
18
20
0 0,2 0,4 0,6 0,8 1
RM
SD
fraction C-gamma exo
89
Figure 53 cis Ac-4,4-F2Pro-OMe in CDCl3.
Figure 54 cis Ac-4,4-F2Pro-OMe in D2O.
6.8. Thermodynamics and kinetics 6.8.1. T1 values
In this section, T1 values of all the molecules in CDCl3 are reported, which makes clear that these are indeed very similar for the cis and trans conformer.
Table 29 T1 values (in s) for all CDCl3 samples at 298 K and 308 K. By integration of 19F signals for Ac-(4R)-FPro-OMe and Ac-(4S)-FPro-OMe or by integration of methyl 1H signals for Ac-Pro-OMe and Ac-4,4-F2Pro-OMe.
Ac-Pro-OMe Ac-(4R)-FPro-OMe Ac-(4S)-FPro-OMe Ac-4,4-F2Pro-OMe
T1,trans 298 K 2.7308 ± 0.0492 2.1890 ± 0.0024 2.1142 ± 0.0058 2.2335 ± 0.0260
T1,cis 298 K 2.7547 ± 0.0582 2.2270 ± 0.0099 2.0087 ± 0.0086 2.1625 ± 0.0457
T1,trans 308 K 3.0344 ± 0.0357 2.6118 ± 0.0030 2.4604 ± 0.0109 2.6558 ± 0.0435
T1,cis 308 K 3.0373 ± 0.0544 2.6468 ± 0.0079 2.3752 ± 0.0151 2.6688 ± 0.0305
0
2
4
6
8
10
12
14
16
18
0 0,2 0,4 0,6 0,8 1
RM
SD
fraction C-gamma exo
0
2
4
6
8
10
12
14
16
0 0,2 0,4 0,6 0,8 1
RM
SD
fraction C-gamma exo
90
6.8.2. Linear regression for determining ΔH° and ΔS°
6.8.2.1. CDCl3 samples
Figure 55 RT ln(K) as a function of temperature for Ac-Pro-OMe in CDCl3.
Figure 56 RT ln(K) as a function of temperature for Ac-(4R)-FPro-OMe in CDCl3.
y = 5,9358x - 5143,8R² = 0,9611
-3480
-3460
-3440
-3420
-3400
-3380
-3360
-3340
-3320
-3300
280 285 290 295 300 305 310R
T ln
(K)
T (K)
y = 5,9344x - 5304,8R² = 0,9358
-3620
-3600
-3580
-3560
-3540
-3520
-3500
-3480
-3460
285 290 295 300 305 310
RT
ln(K
)
T (K)
91
Figure 57 RT ln(K) as a function of temperature for Ac-(4S)-FPro-OMe in CDCl3.
Figure 58 RT ln(K) as a function of temperature for Ac-4,4-F2Pro-OMe in CDCl3.
y = 4,0456x - 2528,8R² = 0,98
-1400
-1380
-1360
-1340
-1320
-1300
-1280
-1260
280 285 290 295 300 305 310
RT
ln(K
)
T (K)
y = 3,5415x - 3797,7R² = 0,9586
-2810
-2800
-2790
-2780
-2770
-2760
-2750
-2740
-2730
-2720
280 285 290 295 300 305
RT
ln(K
)
T (K)
92
6.8.2.2. D2O samples
Figure 59 RT ln(K) as a function of temperature for Ac-Pro-OMe in D2O.
Figure 60 RT ln(K) as a function of temperature for Ac-(4R)-FPro-OMe in D2O.
y = 4,547x - 5383,3R² = 0,9074
-4040
-4020
-4000
-3980
-3960
-3940
-3920
-3900
-3880
-3860
-3840
295 300 305 310 315 320 325 330 335
RT
ln(K
)
T (K)
y = 9,5518x - 7733,3R² = 0,9886
-4950
-4900
-4850
-4800
-4750
-4700
-4650
-4600
-4550
-4500
295 300 305 310 315 320 325 330 335
RT
ln(K
)
T (K)
93
.
Figure 61 RT ln(K) as a function of temperature for Ac-(4S)-FPro-OMe in D2O.
Figure 62 RT ln(K) as a function of temperature for Ac-4,4-F2Pro-OMe in D2O.
y = 3,1891x - 3226,7R² = 0,9978
-2300
-2280
-2260
-2240
-2220
-2200
-2180
-2160
-2140
295 300 305 310 315 320 325 330 335
RT
ln(K
)
T (K)
y = 1,2892x - 3490,2R² = 0,8874
-3120
-3110
-3100
-3090
-3080
-3070
-3060
-3050
295 300 305 310 315 320 325 330 335
RT
ln(K
)
T (K)
94
6.8.3. Linear regression for determining Ea and A (Arrhenius equation)
Figure 63 ln(k1) as a function of 1/RT (blue) and ln(k-1) as a function of 1/RT (orange) for Ac-Pro-OMe in CDCl3.
Figure 64 ln(k1) as a function of 1/RT (blue) and ln(k-1) as a function of 1/RT (orange) for Ac-(4R)-FPro-OMe in CDCl3.
y = -92467x + 34,771R² = 0,9996
y = -95519x + 34,642R² = 0,9994
-4,5
-4
-3,5
-3
-2,5
-2
-1,5
-1
-0,5
0
0,0003880,000390,0003920,0003940,0003960,000398 0,0004 0,0004020,0004040,000406
ln(k
)
1/RT
y = -82005x + 31,484R² = 0,9999
y = -84503x + 31,069R² = 0,9993
-3,5
-3
-2,5
-2
-1,5
-1
-0,5
0
0,0003880,000390,0003920,0003940,0003960,000398 0,0004 0,0004020,0004040,000406
ln(k
)
1/RT
95
Figure 65 ln(k1) as a function of 1/RT (blue) and ln(k-1) as a function of 1/RT (orange) for Ac-(4S)-FPro-OMe in CDCl3.
Figure 66 ln(k1) as a function of 1/RT (blue) and ln(k-1) as a function of 1/RT (orange) for Ac-4,4-F2Pro-OMe in CDCl3.
y = -98899x + 37,892R² = 0,9956
y = -99692x + 37,706R² = 0,9928
-3
-2,5
-2
-1,5
-1
-0,5
0
0,0003880,000390,0003920,0003940,0003960,000398 0,0004 0,0004020,0004040,000406
ln(k
)
1/RT
y = -92555x + 36,791R² = 0,995
y = -95974x + 37,063R² = 0,9961
-2
-1,5
-1
-0,5
0
0,5
1
0,0003880,000390,0003920,0003940,0003960,000398 0,0004 0,0004020,0004040,000406
ln(k
)
1/RT
96
6.8.4. Linear regression for determining ΔH‡ and ΔS‡ (Eyring equation)
Figure 67 R ln(hk1/kBT) as a function of 1/T (blue) and R ln(hk-1/kBT) as a function of 1/T (orange) for Ac-Pro-OMe in CDCl3.
Figure 68 R ln(hk1/kBT) as a function of 1/T (blue) and R ln(hk-1/kBT) as a function of 1/T (orange) for Ac-(4R)-FPro-OMe in CDCl3.
y = -89948x + 35,734R² = 0,9996
y = -93001x + 34,661R² = 0,9994
-280
-275
-270
-265
-260
-255
0,00324 0,00326 0,00328 0,0033 0,00332 0,00334 0,00336 0,00338
Rln
(hk/
kBT)
1/T
y = -79487x + 8,4114R² = 0,9999
y = -81984x + 4,9561R² = 0,9992
-275
-270
-265
-260
-255
-250
-245
0,00324 0,00326 0,00328 0,0033 0,00332 0,00334 0,00336 0,00338
Rln
(hk/
kBT)
1/T
97
Figure 69 R ln(hk1/kBT) as a function of 1/T (blue) and R ln(hk-1/kBT) as a function of 1/T (orange) for Ac-(4S)-FPro-OMe in
CDCl3.
Figure 70 R ln(hk1/kBT) as a function of 1/T (blue) and R ln(hk-1/kBT) as a function of 1/T (orange) for Ac-4,4-F2Pro-OMe in CDCl3.
6.9. Spectra
In the following section all 1D 1H (without 19F decoupling), 1D 19F (with 1H decoupling)
and 2D 1H-13C HSQC spectra for all samples can be found.
y = -96380x + 61,688R² = 0,9953
y = -97173x + 60,138R² = 0,9924
-268
-266
-264
-262
-260
-258
-256
-254
-252
-250
0,00324 0,00326 0,00328 0,0033 0,00332 0,00334 0,00336 0,00338
Rln
(hk/
kBT)
1/T
y = -90036x + 52,528R² = 0,9947
y = -93455x + 54,796R² = 0,9959
-260
-255
-250
-245
-240
-235
0,00324 0,00326 0,00328 0,0033 0,00332 0,00334 0,00336 0,00338
Rln
(hk/
kBT)
1/T
98
Figure 71 1D 1H spectrum without 19F decoupling of Ac-Pro-OMe in CDCl3. 298 K. 700 MHz.
99
Figure 72 1D 1H spectrum without 19F decoupling of Ac-(4R)-FPro-OMe in CDCl3. 298 K. 500 MHz.
100
Figure 73 1D 1H spectrum without 19F decoupling of Ac-(4S)-FPro-OMe in CDCl3. 298 K. 700 MHz.
101
Figure 74 1D 1H spectrum without 19F decoupling of Ac-4,4-F2Pro-OMe in CDCl3. 298 K. 500 MHz.
102
Figure 75 1D 1H spectrum without 19F decoupling of Ac-Pro-OMe in D2O. 298 K. 500 MHz.
103
Figure 76 1D 1H spectrum without 19F decoupling of Ac-(4R)-FPro-OMe in D2O. 298 K. 500 MHz.
104
Figure 77 1D 1H spectrum without 19F decoupling of Ac-(4S)-FPro-OMe in D2O. 298 K. 500 MHz.
105
Figure 78 1D 1H spectrum without 19F decoupling of Ac-4,4-F2Pro-OMe in D2O. 298 K. 500 MHz.
106
Figure 79 1D 19F spectrum with 1H decoupling of Ac-(4R)-FPro-OMe. Left: CDCl3. Right: D2O. Both at 298 K and 500 MHz.
Figure 80 1D 19F spectrum with 1H decoupling of Ac-(4S)-FPro-OMe. Left: CDCl3. Right: D2O. Both at 298 K and 500 MHz.
107
Figure 81 1D 19F spectrum with 1H decoupling of Ac-4,4-F2Pro-OMe in CDCl3. 298 K. 500 MHz.
Figure 82 1D 19F spectrum with 1H decoupling of Ac-4,4-F2Pro-OMe in D2O. 298 K. 400 MHz.
108
Figure 83 2D 1H-13C HSQC of Ac-Pro-OMe in CDCl3. 298 K. 700 MHz.
Figure 84 2D 1H-13C HSQC of Ac-(4R)-FPro-OMe in CDCl3. 298 K. 500 MHz.
109
Figure 85 2D 1H-13C HSQC of Ac-(4S)-FPro-OMe in CDCl3. 298 K. 700 MHz.
Figure 86 2D 1H-13C HSQC of Ac-4,4-F2Pro-OMe in CDCl3. 298 K. 500 MHz.
110
Figure 87 2D 1H-13C HSQC of Ac-Pro-OMe in D2O. 298 K. 500 MHz.
Figure 88 2D 1H-13C HSQC of Ac-(4R)-FPro-OMe in D2O. 298 K. 500 MHz.
111
Figure 89 2D 1H-13C HSQC of Ac-(4S)-FPro-OMe in D2O. 298 K. 700 MHz.
Figure 90 2D 1H-13C HSQC of Ac-4,4-F2Pro-OMe in D2O. 298 K. 500 MHz.
112
7. References
1. Salwiczek, M., et al., Fluorinated amino acids: compatibility with native protein structures and effects on protein-protein interactions. Chem Soc Rev, 2012. 41(6): p. 2135-71.
2. Thomas, K.M., et al., Proline editing: a divergent strategy for the synthesis of conformationally diverse peptides. Org Lett, 2005. 7(12): p. 2397-400.
3. Milner-White, E.J., L.H. Bell, and P.H. Maccallum, Pyrrolidine ring puckering in cis and trans-proline residues in proteins and polypeptides. J Mol Biol, 1992. 228(3): p. 725-34.
4. Renner, C., et al., Fluoroprolines as tools for protein design and engineering. Angew Chem Int Ed Engl, 2001. 40(5): p. 923-5.
5. Bretscher, L.E., et al., Conformational stability of collagen relies on a stereoelectronic effect. J Am Chem Soc, 2001. 123(4): p. 777-8.
6. Newberry, R.W., et al., Signatures of n-->pi* interactions in proteins. Protein Sci, 2014. 23(3): p. 284-8.
7. Newberry, R.W., et al., n-->pi* interactions of amides and thioamides: implications for protein stability. J Am Chem Soc, 2013. 135(21): p. 7843-6.
8. DeRider, M.L., et al., Collagen stability: insights from NMR spectroscopic and hybrid density functional computational investigations of the effect of electronegative substituents on prolyl ring conformations. J Am Chem Soc, 2002. 124(11): p. 2497-505.
9. Hinderaker, M.P. and R.T. Raines, An electronic effect on protein structure. Protein Sci, 2003. 12(6): p. 1188-94.
10. Adzhubei, A.A., M.J. Sternberg, and A.A. Makarov, Polyproline-II helix in proteins: structure and function. J Mol Biol, 2013. 425(12): p. 2100-32.
11. Horng, J.C. and R.T. Raines, Stereoelectronic effects on polyproline conformation. Protein Sci, 2006. 15(1): p. 74-83.
12. Moradi, M., et al., Conformations and free energy landscapes of polyproline peptides. Proc Natl Acad Sci U S A, 2009. 106(49): p. 20746-51.
13. Kay, B.K., M.P. Williamson, and M. Sudol, The importance of being proline: the interaction of proline-rich motifs in signaling proteins with their cognate domains. FASEB J, 2000. 14: p. 231-41.
14. Best, R.B., et al., Effect of flexibility and cis residues in single-molecule FRET studies of polyproline. Proc Natl Acad Sci U S A, 2007. 104(48): p. 18964-9.
15. Doose, S., et al., Probing polyproline structure and dynamics by photoinduced electron transfer provides evidence for deviations from a regular polyproline type II helix. Proc Natl Acad Sci U S A, 2007. 104(44): p. 17400-5.
16. Bochicchio, B. and A. Pepe, Role of polyproline II conformation in human tropoelastin structure. Chirality, 2011. 23(9): p. 694-702.
17. O'Hagan, D., Understanding organofluorine chemistry. An introduction to the C-F bond. Chem Soc Rev, 2008. 37(2): p. 308-19.
18. Odar, C., M. Winkler, and B. Wiltschi, Fluoro amino acids: a rarity in nature, yet a prospect for protein engineering. Biotechnol J, 2015. 10(3): p. 427-46.
19. O'Hagan, D. and H. Deng, Enzymatic fluorination and biotechnological developments of the fluorinase. Chem Rev, 2015. 115(2): p. 634-49.
20. Buer, B.C. and E.N. Marsh, Fluorine: a new element in protein design. Protein Sci, 2012. 21(4): p. 453-62.
113
21. Sharaf, N.G. and A.M. Gronenborn, (19)F-modified proteins and (19)F-containing ligands as tools in solution NMR studies of protein interactions. Methods Enzymol, 2015. 565: p. 67-95.
22. Panasik, N., et al., Inductive effects on the structure of proline residues. International Journal of Peptide and Protein Research, 1994. 44(3): p. 262-269.
23. Hodges, J.A. and R.T. Raines, Stereoelectronic and steric effects in the collagen triple helix: toward a code for strand association. J Am Chem Soc, 2005. 127(45): p. 15923-32.
24. Kim, W., K.I. Hardcastle, and V.P. Conticello, Fluoroproline flip-flop: regiochemical reversal of a stereoelectronic effect on peptide and protein structures. Angew Chem Int Ed Engl, 2006. 45(48): p. 8141-5.
25. Shi, G.-Q. and W.-L. Cai, d,e-unsaturated b,b-difluoro-a-keto esters: novel synthesis and utility as precursors of b,b-difluoro-a-amino acids. J Org Chem, 1995. 60(20): p. 6289-95.
26. Liu, Z., et al., 3-Fluoroazetidinecarboxylic Acids and trans,trans-3,4-Difluoroproline as Peptide Scaffolds: Inhibition of Pancreatic Cancer Cell Growth by a Fluoroazetidine Iminosugar. J Org Chem, 2015. 80(9): p. 4244-58.
27. Hommel, U., et al., Pyrrolidine derivatives and their use as complement pathway modulators. 2014, Google Patents.
28. Wüthrich, K., NMR of proteins and nucleic acids. 1986. 29. Zangger, K., Pure shift NMR. Prog Nucl Magn Reson Spectrosc, 2015. 86-87: p. 1-20. 30. Foroozandeh, M., et al., Ultrahigh-resolution NMR spectroscopy. Angew Chem Int Ed
Engl, 2014. 53(27): p. 6990-2. 31. Sinnaeve, D., et al., A general method for extracting individual coupling constants from
crowded 1H NMR spectra. Angew Chem Int Ed Engl, 2016. 55: p. 1090-3. 32. Aguilar, J.A., G.A. Morris, and A.M. Kenwright, “Pure shift” 1H NMR, a robust method
for revealing heteronuclear couplings in complex spectra. RSC Advances, 2014. 4(16): p. 8278.
33. Boeyens, J.C.A. and D.G. Evans, Group theory of ring pucker. Acta Cryst, 1989. B45: p. 577-81.
34. Cremer, D. and J.A. Pople, Molecular orbital theory of the electronic structure of organic compounds. XXIII. Pseudorotation in saturated five-membered ring compounds. J Am Chem Soc, 1975. 97(6): p. 1358-67.
35. Altona, C. and M. Sundaralingam, Conformational analysis of the sugar ring in nucleosides and nucleotides. A new description using the concept of pseudorotation. J Am Chem Soc, 1972. 94(23): p. 8205-12.
36. Hendrickx, P.M.S., Conformational analysis of flexible molecules, in Organic Chemistry. 2009, Ghent University.
37. Karplus, M., Contact electron-spin coupling of nuclear magnetic moments. J Chem Phys, 1959. 30: p. 11-5.
38. Postova Slavetinska, L., D. Rejman, and R. Pohl, Pyrrolidine nucleotide analogs with a tunable conformation. Beilstein J Org Chem, 2014. 10: p. 1967-80.
39. Diez, E., J. San-Fabian, and J. Guilleme, Vicinal proton-proton coupling constants I. Formulation of an equation including interactions between substituents. Molecular Physics, 1989. 68(1): p. 49-63.
114
40. Guilleme, J., J. San-Fabian, and E. Diez, Vicinal proton-proton coupling constants II. Analysis of the effect of interaction between geminal substituents upon vicinal couplings to methyl groups. Molecular Physics, 1989. 68(1): p. 65-85.
41. Marshall, J.L., et al., Reasons for the nonequivalence of the exo-exo and endo-endo vicinal NMR coupling constants in norbornanes. Tetrahedron, 1976. 32: p. 537-42.
42. Thibaudeau, C., J. Plavec, and J. Chattopadhyaya, A new generalized Karplus-type equation relating vicinal proton-fluorine coupling constants to H-C-C-F torsion angles. J Org Chem, 1998. 63(15): p. 4967-84.
43. Hendrickx, P.M. and J.C. Martins, A user-friendly Matlab program and GUI for the pseudorotation analysis of saturated five-membered ring systems based on scalar coupling constants. Chem Cent J, 2008. 2: p. 20.
44. Alper, J.S. and R.I. Gelb, Standard errors and confidence intervals in nonlinear regression: comparison of Monte Carlo and parametric statistics. J Phys Chem, 1990. 94(11): p. 4747-51.
45. Hobart, D.B., Jr. and J.S. Merola, (2S,4R)-4-Fluoro-pyrrolidinium-2-carboxyl-ate. Acta Crystallogr Sect E Struct Rep Online, 2012. 68(Pt 8): p. o2490.
46. Siebler, C., et al., Importance of dipole moments and ambient polarity for the conformation of Xaa–Pro moieties – a combined experimental and theoretical study. Chem. Sci., 2015. 6(12): p. 6725-6730.
47. Eberhardt, E.S., et al., Solvent effects on the energetics of prolyl peptide bond isomerization. J Am Chem Soc, 1992. 114(13): p. 5437-9.
48. Eberhardt, E.S., N. Panisik, Jr., and R.T. Raines, Inductive Effects on the Energetics of Prolyl Peptide Bond Isomerization: Implications for Collagen Folding and Stability. J Am Chem Soc, 1996. 118(49): p. 12261-12266.
49. Thomas, C.A., E.R. Talaty, and J.G. Bann, 3S-fluoroproline as a probe to monitor proline isomerization during protein folding by 19F-NMR. Chem Commun (Camb), 2009(23): p. 3366-8.
50. Jacobsen, N.E., NMR spectroscopy explained. 2007. 51. Sinnaeve, D., Gevorderde structuuranalyse van een recent geïsoleerd natuurproduct
met behulp van NMR spectroscopie. 2005, Ghent University. 52. Cory, D.G. and W.M. Ritchey, Suppression of signals from the probe in Bloch decay
spectra. J Magn Reson, 1988. 80: p. 128-32. 53. Thrippleton, M.J. and J. Keeler, Elimination of Zero-Quantum Interference in Two-
Dimensional NMR Spectra. Angewandte Chemie, 2003. 115(33): p. 4068-4071. 54. Kupce, E., J. Boyd, and I.D. Campbell, Short selective pulses for biochemical
applications. J Magn Reson Series B, 1995. 106: p. 300-3. 55. Geen, H. and R. Freeman, Band-selective radiofrequency pulses. J Magn Reson, 1991.
93: p. 93-141.
![M1-pipeline scalaire-1213.ppt [Mode de compatibilité]de/M1-pipeline_scalaire-1213.pdf · 2012-10-26 · M1 Informatique 2012-13 Architectures avancées D. Etiemble 11 Immediate ux](https://img.pdfslide.us/doc/110x75/5f452e481c0a36382d1b85d9/m1-pipeline-scalaire-1213ppt-mode-de-compatibilit-dem1-pipelinescalaire-1213pdf.jpg)
